Français Tirailleur Pidgin

— a corpus study

Hedvig Skirgård

Department of Linguistics, Stockholm UniveristyThesis submitted for Master of Arts in Linguistics(30 HE credits)Spring 2013Supervisor: Mikael ParkvallExaminer: Henrik LiljegrenExpert reviewer: Bernhard Wälchli

Abstract

Swedish

Français Tirailleur (FT) är ett pidginspråk som talades av västafrikanska sol-

dater och deras vita officerare i den franska kolonialarmen cirka 1857-1954. Den här

uppsatser beskriver denna språkvatietet utifrån ett korpus som består av de doku-

menterade yttranden som hittills hittats. Studien visar bland annat att standard

negation uttrycks med en pre-verbal partikel (pas), ja/nej-frågor uttryckts främst

genom intonation, genussystemet är inte produktivt, det finns ingen skillnad mellan

subjekt och objekt i pronomensystemet och attributiv ägande uttrycks med possessiva

pronomen, juxtaposition eller prepositioner. Den standardiserade type-token-ration

är 26%, vidare forskning om ordförråd i pidginspråk och jämförelser med talat språk

behövs. Det finns två former som är väldigt frekventa och som anses vara mycket

karakteristiska för FT: ya och yena. Dessa former har tidigare beskrivits som stativa

verb, relativmarkörer och finithetsmarkörer. De förekommer i majoriteten av alla

dokument i korpuset. De fungerar som stativa verb, kopula eller kopulalika markörer,

samt potentiellt även som predikatsmarkörer. Frågan om huruvida adjektiv är en

relevant språklig kategori i denna pidgin diskuteras också i denna uppsats.

Nyckelord: lingvistik, litet korpus, pidgin, tiraljör, andra världskriget, västafrika,

senegal, petit nègre, moi-ya-dit, anonym manual 1916, Charles Mangin, Lucie Cous-

turier, kolonialism

English

Français Tirailleur (FT) is a pidgin language that was spoken by West African

soldiers and their white officers in the French colonial army approximately 1857-1954.

The aim of this study is to investigate a corpus of previously unanalyzed utterances

of FT in order to discern linguistic structures and test previous statements about the

nature of FT. Much of previous literature on FT is based on an anonymous manual

published by the French military in 1916, this thesis aims to provide new informa-

tion to our understanding of this pidgin. These are some of the findings: standard

negation is expressed by means of a preverbal particle (pas), polar interrogation by

intonation, grammatical gender is not a productive category and attributive posses-

sion is expressed by possessive pronouns, juxtaposition (possessum - possessor) and

prepositional constructions. The standardized type-token-ratio of this corpus, 26%,

suggests that the lexicon of pidgins needs to be further studied. Comparisons with

corpuses of spoken language are needed. There are two very frequent pre-predicate

markers that are considered characteristic of FT: ya and yena. These two markers

have previously been described as stative verbs, relativizers and markers of finiteness.

The two markers are very frequent in a majority of the sources and are highly poly-

semous, functioning as stative verbs, copula or copula-like markers and possibly also

predicate markers. The status of adjectives as a part-of-speech in FT is also discussed.

Keywords: Linguistics, Corpus, Pidgin, Tirailleur, Sénégalais, West Africa, Petit-

Nègre, Moi-Ya-Dit, Charles Mangin, Lucie Cousturier, Dakar 1944, First World War,

WW1, Second World War, WW2, APiCS, Senegal, colonialism

i

Contents

1 Introduction 1

2 Research questions 2

3 Background 33.1 Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1.1 Theory, assumptions and frameworks . . . . . . . . . . . . . . . . . 33.1.2 Pidgins and contact languages . . . . . . . . . . . . . . . . . . . . . 53.1.3 Summary of theoretical background . . . . . . . . . . . . . . . . . . 7

3.2 Les Tirailleurs Sénégalais, West Africa and France . . . . . . . . . . . . . . 73.2.1 Brief account of the history of contact between France and West Africa 83.2.2 The pidgin of Français Tirailleur . . . . . . . . . . . . . . . . . . . 113.2.3 Summary of previous statements about the grammar of FT . . . . . 123.2.4 The linguistic makeup of the Français Tirailleur troops . . . . . . . 163.2.5 Summary: Les Tirailleurs Sénégalais, West Africa and France . . . 18

4 Method 204.1 Source material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204.2 Reliability of the sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234.3 Comparison with other languages . . . . . . . . . . . . . . . . . . . . . . . 234.4 Limitations of the study – the nature of corpus . . . . . . . . . . . . . . . 234.5 Annotation of the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5 Results and analysis 265.1 Data makeup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

5.1.1 Clause-types and stative/dynamic . . . . . . . . . . . . . . . . . . . 265.1.2 Type-token-ratio in the corpus . . . . . . . . . . . . . . . . . . . . . 275.1.3 Summary: data makeup . . . . . . . . . . . . . . . . . . . . . . . . 31

5.2 Function seeking form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.2.1 Polar Interrogative . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.2.2 Standard Negation . . . . . . . . . . . . . . . . . . . . . . . . . . . 345.2.3 Non-standard negation (jamais, rien, personne, plus) . . . . . . . . 375.2.4 Grammatical gender . . . . . . . . . . . . . . . . . . . . . . . . . . 385.2.5 Personal pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . 405.2.6 Attributive possession . . . . . . . . . . . . . . . . . . . . . . . . . 435.2.7 Predicative possession . . . . . . . . . . . . . . . . . . . . . . . . . 455.2.8 Copula relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465.2.9 ‘Being able’ — moyen and pouvoir . . . . . . . . . . . . . . . . . . 475.2.10 Summary: function seeking form . . . . . . . . . . . . . . . . . . . 48

5.3 Form seeking function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485.3.1 Drop of overt subject . . . . . . . . . . . . . . . . . . . . . . . . . . 48

ii

5.3.2 Polysemous form ya . . . . . . . . . . . . . . . . . . . . . . . . . . 515.3.3 Polysemous form yena . . . . . . . . . . . . . . . . . . . . . . . . . 555.3.4 Gagner — to get, become or have? . . . . . . . . . . . . . . . . . . 565.3.5 Wanting, liking and being happy — content and vouloir . . . . . . 575.3.6 Reduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575.3.7 Summary: form seeking function . . . . . . . . . . . . . . . . . . . 59

5.4 Notes on diachrony . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.4.1 Order of modifier and head . . . . . . . . . . . . . . . . . . . . . . 605.4.2 Aspectual fini? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.4.3 Pre-verbal qui/que . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.4.4 Summary: notes on diachrony . . . . . . . . . . . . . . . . . . . . . 61

6 Discussion 62

7 Conclusions 64

iii

List of Figures

1 Map of European expansion in Africa in 1914 (Guillaume Balavoine 2013)after the Scramble for Africa c�2002 Guillaume Balavoine. . . . . . . . . . 9

2 Cover of the anonymous manual of FT. c�1916 Imprimerie Militaire Universelle 113 Cover of La Force Noir by Mangin (1910). c�1910 Charles Mangin. . . . . 164 Cover of Épopées Africaines by Baratier (1912) featuring an illustration by

Lucien Pouzargues. c�1912 Albert Baratier and Lucien Pouzargues. . . . . 225 Type-frequency-ratio of the FT-corpus. X-axis = types ranked by freq, Y-

axis=frequency. Non-Logarithmic. . . . . . . . . . . . . . . . . . . . . . . . 296 Type-frequency-ratio of the FT-corpus. Non-logarithmic, zoomed in at first

100 types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 Type-frequency of the FT-corpus. X-axis = types ranked by freq, Y-axis=frequency

(log-log=10) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Advertisement for chocolate powder. c�1915 Giacomo de Andreis. . . . . . 51

List of Tables

1 Brief historical overview of important events in the history of Les TirailleurSénégalais . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2 Languages proposed as involved in the multilingual situation of FT. . . . . 183 FT material sorted by year of production of utterance . . . . . . . . . . . . 214 The eight largest documents of the corpus . . . . . . . . . . . . . . . . . . 215 Independent/subordinate clauses . . . . . . . . . . . . . . . . . . . . . . . 266 Declarative/interrogative/imperative/other . . . . . . . . . . . . . . . . . . 267 Stative and dynamic clauses . . . . . . . . . . . . . . . . . . . . . . . . . . 278 Types/tokens in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289 Polar interrogatives in certain West African languages . . . . . . . . . . . . 3210 Expression of Standard Negation based on Dryer (2011c,i,e). . . . . . . . . 3511 Expression of Standard Negation in FT . . . . . . . . . . . . . . . . . . . . 3512 Expression of non-standard Negation in FT . . . . . . . . . . . . . . . . . 3813 Gender marking in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3914 French pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4115 Pronouns in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4116 French possessive pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . 4317 Expression of attributive possession in FT . . . . . . . . . . . . . . . . . . 4418 Predicate possession in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . 4519 Copula relations distinguished in this study . . . . . . . . . . . . . . . . . 4620 Expression of copula relations in FT . . . . . . . . . . . . . . . . . . . . . 4721 Drop of overt subject . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4922 Null subject distributed over person and number . . . . . . . . . . . . . . . 5023 Ya in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

iv

24 Ya besoin and ya moyen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5425 Yena in FT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5526 Loans with reduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

List of glosses

1sg 1st person singular2sg 2nd person singular3sg 3rd person singular1pl 1st person plural2pl 2nd person plural3pl 3rd person pluralACC Accusative caseAP Adjective PhraseCOP CopulaDAT Dative caseDEF Definite articleDEM DemonstrativeDN Double NegationFUT Future tenseINDEF Indefinite articleNEG NegationNOM Nominative caseNP Nominal PhrasePASS PassivePOSS PossessivePP Prepositional PhrasePROG Progressive aspectPRS Present tensePro PronounPST Past tense (“Simple” Past)PTCP ParticipleSN Standard NegationYa the auxiliary/stative verb/predicate marker ya. 1

Yena the auxiliary/stative verb/predicate marker yena

v

List of abbreviations

APiCS Atlas of Pidgin and Creole Structures (see Michaelis et al. 2013)BLT Basic Linguistic Theory (see Dixon 2010a)ELICOP Étude Linguistique de la Communication Parlée (see Debrock et al. 2001)FT Français TirailleurTTR Type-token-ratio. The number of types divided by the number of tokens,

usually displayed in procent.sTTR Standardized type-token-ratio (only of 2 000 consecutive words)2

SIL Summer Institute of Linguistics - faith-based nonprofit organizationfounded in 1934. SIL is devoted to the description, documentation andpreservation of the world’s languages.

stdzd Standardized materialWALS Word Atlas of Language Structures (see Dryer and Haspelmath 2011)WW1 First World WarWW2 Second World War

Format of examples from corpus

Each example from the corpus contains a reference-ID that identifies the entry in thecorpus (ex. FrTir0286b) and a reference to the original source. Please note that the date ofutterance is not necessarily the same as the date of publication of document. The examplesare given in four lines.

Original The string exactly as in the sourceStdzd The string converted to a more standardized spelling as to make the

corpus more coherent, comparable and searchable (by Mikael Parkvall)Gloss Glossing of the stdzd-string (by Mikael Parkvall)Translation Translation into English, either by the source or Mikael Parkvall

Example:

Français Tirailleur (FrTir0284, Anon 1916:54)Original Ramener jambe droite en avant, genou droite bien fléchi

Stdzdgloss

Ramenerbring

jambeleg

droiteright

enavant,forward,

genouknee

droiteright

bienwell/very

fléchibent

’Put your right leg forward, with your knee bent’

vi

1 Introduction

This thesis is a descriptive study of a contact language known as Français Tirailleur (FT)based on corpus material.

When speakers of different languages have to communicate despite a lack of commonlanguage several things can happen; they can settle on one of the involved languages asa common tongue, but they can also create a new language variety. These languages arecalled contact languages, and this thesis is devoted to the description of one such languagevariety, FT.

FT is a contact language with no mother tongue speakers — a pidgin. It was spokenby West African soldiers and their white officers, approximately between 1857-1954. TheseWest African soldiers were enrolled in the French Colonial army and had to communicate,not only with superior white French officers but also with the fellow soldiers with whomthey did not necessarily share a common language. The sentence in (1) was uttered bya West African soldier, a Tirailleur Sénégalais, in the 1940’s and illustrates the situationquite well.

(1) Français Tirailleur (FrTir0377, Woodfork 2001:108)Original Oui, parce que lui parler sa langue, je comprends pas, et moi parler malangue, lui comprend pas

StdzdGloss

Oui,yes,

parce.quebecause

lui3sg

parlerspeak

sa3sg.POSS

langue,language

je1sg

comprendsunderstand

pas,NEG,

etand

moi1sg

parlerspeak

ma1sg.POSS

langue,language,

lui3sg

comprendunderstand

pasNEG

‘Yes, because if he speaks his language, I don’t understand, and if I speak mine, hewon’t understand’

There are several different types of contact language varieties and there are also manydifferent definitions of these varieties. As Muysken and Smith (1995:3) writes “[c]reolistsagree neither about the precise definition of the terms pidgin and creole, nor about thestatus of a number of languages that have been claimed to be pidgins and creoles”. Onedefinition of a contact language that is often cited is that by Thomason (1997:3): “alanguage that arises as a direct result of language contact and that comprises linguisticmaterial which cannot be traced back primarily to a single source language”.

While there are many different theories concerning the nature and origin of contactlanguages, one distinction that many scholars do agree on is that there are contact languagevarieties that have native speakers and there are those that do not. This thesis is concernedwith the latter and specifically languages that are stable enough to be labeled ‘pidgins’.The subject matter of what is and what is not a creole and how they came about, whilehighly interesting and worthy of more debate, is not discussed in this thesis.

Contact languages usually include elements from many different languages, but the bulk

1

of the vocabulary is often from one language — this is what is known as the lexifier (orsuperstrate) language. Most languages of the world contain words and elements from otherlanguages, what makes pidgins in particular different is not the mix of linguistic materialper se, but rather the reduction of grammar and lexicon. Pidgins have been noted to arisein situations of great need such as war and slave trade but also through trade (Foley 2006).

The language variety that we are concerned with in this thesis goes by several names,among others: Forofifon Naspa, Moi- ya-dit, Petit-Nègre and Français Tirailleur. I will inthis thesis use the term Français Tirailleur (FT). The word tirailleur in French translatesto infantryman or rifleman in English. African soldiers from Algeria and Morocco werealso called ‘Tirailleurs’, this thesis will however only concern Tirailleurs of West Africa,Les Tirailleur Sénégalais.

There is one source on FT that has been cited more often than any other, the anonymousmanual from 1916 (Le français tel que le parlent nos tirailleurs Sénégalais). This thesisaims to provide more insights into the nature of FT, using not only material from thewell-cited anonymous manual but from many more documents and authors.

The corpus of FT consists of material from 148 different documents, 130 authors, 8686 words and 1737 strings. This material has been collected, glossed and translated bycontact language researcher Mikael Parkvall and annotated by me.

2 Research questions

The aim of this thesis is to investigate the structure of Français Tirailleur (FT) through asmall corpus. The major questions are

i What grammatical features can we discern from this corpus?

ii How does FT differ from the lexifier French with respect to those features?,

iii Is there any evidence of influence from West African languages with respect to thosefeatures?,

iv Is there evidence of change over time?

Much of the literature on FT has focused on the anonymous manual from 1916, one ofthe aims of this thesis is to test if there is support for the statements that have been madepreviously in earlier literature.

2

3 Background

The background section is divided into two parts. The first part concerns the theoreticalbackground of the study and the second part provides a background for the multilingualsituation where this variety arose.

3.1 Theoretical background

3.1.1 Theory, assumptions and frameworks

A language variety is a conventionalized system for expressing meaning between humans.Languages are in their nature fluid, both synchronically and diachronically. This meansthat the task of a linguist (understanding human language) is very hard indeed. There is afine line between what we define as two separate languages and what we label as dialects ofone language. Contact language varieties are by no means not exempt from this. What iscalled one language variety can, usually, also be described as a collection different varieties,perhaps overlapping in certain areas more than others. This is the reality of a linguist’sobject of study and something we always need to keep in mind.

A problem of research on pidgins in particular is that the object of study is moreunstable in their structure than languages on average. It is necessary to take this variationinto account and recognize the limitations of what we can understand and describe in alanguage. This is the reason why the focus of this study is on basic structures which weexpect to be more stable in this pidgin variety and more frequent in our corpus, such asnegation, polar interrogation and attributive possession.

Theoretical linguistics have suggested that this variation and change is not random,there are patterns in this variation; some features might be dependent on others andcertain paths of language change might be more common than others.

I will in this thesis make use of the theoretical concepts, assumptions and framework(s)as found in much of descriptive linguistics and linguistic typology, such as WALS (Dryerand Haspelmath 2011). Many general linguists might work outside specific theories andframeworks, such as Minimalism or Role and Reference Grammar, but there are nonethelessstill certain theoretical assumptions present.

Many of the assumptions that are widely used in language description and linguistictypology are summarized in the works on Basic Linguistic Theory (BLT) by R.M.W Dixonand Matthew Dryer. BLT aims at providing a cumulative framework of descriptive lin-guistics, not an explanatory one. BLT contains many concepts that are widespread andused by many different scholars in the field of theoretical linguistics.

Ideally, descriptive linguists and typologist should work with descriptions of languagethat are the result of an analysis of those language on their own terms and not basedon knowledge of categories in other languages. This is what Haspelmath (2010b) calls“framework-free”; descriptive linguists should employ a set of concepts that are constructedseparately for each language. This is, sadly, seldom feasible, and in the case of FT mostlikely impossible. We cannot gather new data, and what is more relevant: we have little

3

or no negative evidence illustrating what is ungrammatical.How are we then to do cross-linguistic comparisons at all? Linguists use labels that

are intended to cover a certain shared function/meaning — cf. Haspelmath’s ComparativeConcepts, the gram-types of the ‘Bybee&Dahl-approach’ (Bybee and Dahl 1989) or thelabels of BLT. These concepts are useful for linguists as they facilitate communicationbetweens scholars and cross-linguistic comparisons, but we should always keep in mindthat this does not mean that these categories are the same across all languages.

Many scholars have emphasized the need to distinguish between language-specific de-scriptive categories and comparative categories intended for cross-linguistic comparison.Some researchers, such as (Haspelmath 2010a), have proposed that these comparativecategories are created for the benefit of linguistic typology alone and need not have anybearings on reality. Others, such as Dahl (p.c.), suggest that these cross-linguistic cate-gories (‘gram-types’) might represent probabilistic clusters of linguistic function that havepotential bearing on the reality of speakers’ minds.

FT is most likely an unstable language variety or collection of varieties,; as are most(if not all) pidgins. FT is probably also a dead language variety (though this needs to bemade certain) and the written records are very limited. The modest aim of this thesis isto discern what structure is possible to grasp from the corpus material, focusing on high-frequent phenomena and grammatical structures that are believed to be more stable, suchas negation, polar interrogation, possession and the expression of a formal subject.

In order to investigate these features, I will employ concepts and labels that are wide-spread in language description and in cross-linguistic comparison. The alternative wouldbe to construct new categories based on FT alone. It is, unfortunately, hard to conceivehow that could be accomplished using the limited material available.

Furthermore, I make the theoretical assumption that there is such a thing as a con-ventionalized contact variety (or collection of varieties with considerable overlap in certainareas) with no mother tongue speakers. If it is stable enough I call it a ‘pidgin’ (as op-posed to the more unstable kind that we call ‘jargon’). These language varieties are likelyto exhibit great internal variation. Whether or not creoles originate from these varietiesor not is not discussed here.

4

3.1.2 Pidgins and contact languages

There are several potential outcomes of situations when speakers lacking a common lan-guage need to communicate. These are some of the outcomes linguists have found.

Language shift One or more groups adopt the language of one of the groups. This mightlead to language death for the other languages involved.

Multilingualism Speakers learn the languages of the other group(s) while maintainingtheir native languages, perhaps using the different language for different functionaldomains.

Pidgin A language variety with no mother tongue speakers and restricted grammar andlexicon, primarily used as a between-groups language .

Creole Contact language variety with mother tongue speakers. There is great controversyas to the origin of creoles, that topic will not be discussed here.

Mixed language Mix of two (or more) languages resulting in a variety that shows positivegenetic relationship with more than one language (Bakker 1997:195)3. Not to beconfused with creoles, jargons or pidgins.

A pidgin is a type of contact language with no native speakers. Pidgins are characterizedby reduced grammar and lexicon. Thomason (1997:76) writes that a pidgin prototypically“arises in a new contact situation in which three or more groups of speakers come togetherfor purposes of trade or other limited communicative purposes”.

Thomason’s definition excludes a purely bilingual contact situation, according to her itis the three-language-scenario or more that is key. There are counter examples of this, suchas Russenorsk (Broch and Jahr 1981) — a pidgin formed in a bilingual contact situationof Russian and Norwegian. In the case of FT, however, there is no doubt that the WestAfrican group spoke several different languages resulting in a multilingual situation of atleast more than three languages.

In his book La Force Noir Mangin (1910) writes that the largest ethnical groups amongthe French West African colonial troops were Wolof, Fula 4, Hausa and Mande, see section3.2.4 for more details. These groups of native West Africans did not seem to know eachother’s languages very well, nor did they, typically, master French — the language spoken bytheir commanding officers 5. This means that FT was not only necessary for communicationbetween the French officers and the West African soldiers but also among the soldiersthemselves (see example 1 on page 1). It is not likely though that FT was used between

3Mixed languages typically show similarities with two (or more) in their grammar as well as theirlexicon, a language that draws only lexicon from one language and only grammar from another is usuallynot labeled “mixed”

4This group is also known as Pulaar, Fulfulde and Peul.5Though there are a few exceptions of West Africans who did mastered French.

5

speakers in these troops that did know each others language, FT was in other word abetween-groups-language rather than a in-group-language.

Pidgins tend to be quite unstable and there are very few descriptive works of pidgins.We do not know much about pidgins and it is hard to say how many pidgins there are orhave been. Pidgins are not clearly defined, where do we draw the line between interlan-guage, foreigner talk, jargon, pidgins and creoles?

Holm made an important contribution to the field of contact linguistics with his two-volume survey of pidgins, semi-creoles and creoles that was published in 1988.The twovolumes contain 88 contact languages. Ethnologue - catalogue of the world’s languagesdistributed by SIL - Lewis et al. (2013) contains 17 pidgins (and 93 creoles). Glottolog(Nordhoff et al. 2012) another catalogue over language varieties, contains 35 pidgins. Thesame site that hosts Glottolog also contains Langdoc, a catalogue of language descriptions.Langdoc contains 51 entries of grammars or grammar sketches of pidgins.

The World Atlas of Language Structures (WALS) and The Atlas of Pidgin and CreoleLanguage Structures (APiCS) do not make an explicit distinction between pidgins andcreoles (or any other type of contact language). WALS contains 32 pidgins and creoles(Dryer 2011b) and APiCS encompasses 76 pidgins and creoles (Michaelis et al. 2013).

Pidgins emerge from language contact, but if the groups have too much contact thespeakers might switch to one of the involved languages or the pidgin might expand into acreole (depending on your analysis of the origin of creoles). Holm (1988:4-5) includes thisin his definition of a pidgin “no groups learns the native language of any other group forsocial reasons that may include lack of trust or of close contact”. However, if the groupshave too little contact the language variety might be less stable and consistent — not adistinct language variety at all.

More than 230 language varieties have at least once been defined as pidgins or potentialpidgins by expert scholars (Parkvall p.c.). Of the language varieties that are included inWALS and APiCS there are 6 and 16 respectively that have at least once been labeled“pidgin”. It is important to note that neither WALS nor APiCS claim to include all pidginsor creoles.

Why are there then so few descriptions of pidgins? There are many possible reasons

Not enough research, yet The study of contact language varieties is itself a very youngdiscipline

Too little data There are very few documented pidgin utterances seeing as it is not alanguage variety that is often expressed in writing.

Lack of knowledge Speakers of pidgins are not always aware of the fact that they speaka distinct language variety

Shame Even if they are aware that they master a pidgin language variety it might beconnected to social stigma — motivation from the speakers in describing the varietyis very low

6

Lack of in-group identity Connected to the two previous points; pidgins have no nativespeakers. The speakers all belong to other groups and are perhaps not likely to seethis group in which the pidgin arose as important enough compared to their nativegroup, again making motivation from the speakers in describing the variety is verylow.

Many of the problems in pidgin studies are similar to problems faced by scholars study-ing critically endangered or dead languages.

As previously stated, we as linguists know little about the nature of pidgins, but whatwe do know is that they seem to arise in situations of attenuated contact such as colonialism,slave trade, war or trade (Foley 2006:2). There is no time to learn each other’s languagesthoroughly and the need to communicate seems to be stronger than the desire to belongto a certain group6, the motivation behind pidgins is perhaps what sets them apart fromsecond-language-acquisition — further research on these matters is however needed.

3.1.3 Summary of theoretical background

Ideally, one would describe each language on its own terms — using bottom-up generatedcategories. This is however not feasible in the case of FT, this thesis will instead makeuse of concepts and categorizes that have been used in linguistic typology and languagedescription.

Pidgins are contact-varieties that are characterized by reduced grammar and lexiconand the lack of mother tongue speakers. Pidgins are shaped by a multilingual situationwhere several groups need to communicate, but lack a common language. These situationsneed not produce pidgins, there are other possible outcomes. We know little of the natureof pidgins.

3.2 Les Tirailleurs Sénégalais, West Africa and France

This section contains a background of the contact between France and West Africa and ofthe West African troops, Les Tirailleur Sénégalais.

The first West African soldiers were enrolled in the French colonial army in 1820 andthe company of the Tirailleur Sénégalais was formed in 1857 (Echenberg 1986:311-315).They fought for France in both World Wars. The last company of West African soldiers inthe French army was disbanded in 1964 and the last Tirailleur who served in WW1 diedin 1998 (Michel 2003).

It is important to keep in mind that while slavery was abolished in France and hercolonies in 1848, this does not mean situation changed totally over night. There is aninter-departmental report from 1950 suggesting that the recruitment was not necessaryvoluntary from that point forward. The French military was to go “up-country to enrollcaptives, to whom the sum needed to purchase their freedom is given as enrollment bounty”

6The group being native or native-like speakers.

7

(Hargreaves, 1969: 100 as cited in Wilson (1999:10)). In other words, they more or lessbought captives and turned them into soldiers.

3.2.1 Brief account of the history of contact between France and West Africa

There is evidence of French presence in West Africa since the 16th century. The Frenchcolonized West Africa, extracting slaves, gold and other goods. The English, Spanish,Portuguese and to a less extent the German, Danish and Dutch were also present in thearea.

On the following page is a map (figure 1) illustrating the entire African continent of 1914(Guillaume Balavoine 2013). This map shows the partition of Africa after the Scramblefor Africa of 1881- 1914. The Scramble of Africa is a name given to a intense period ofinvasion, colonization and annexation of the africa continent by European colonial powers.France was at that time in possession of great parts of West Africa (Afrique OccidentaleFrançaise), Equatorial Africa and the entire island of Madagascar. Liberia and Abyssinia(present day northern Ethiopia and Eritrea) were independent. The states of West Africawere all independent by the end of the 1960’s (Chafer 2002).

At the time of the French colonization of West Africa many Africans were labelled as“Sénégalais” even if they were not from Senegal. The is also reflected in the name of theWest African troops in the French colonial army; Les Tirailleur Sénégalais.

Table 1 contains a brief overview of some of the major events in the history of the contactbetween France and West Africa. For a more extensive account, see (Wilson 1999:3-21).

8

Figure 1: Map of European expansion in Africa in 1914 (Guillaume Balavoine 2013) afterthe Scramble for Africa c�2002 Guillaume Balavoine.

9

time period source1340-1470 European expansion along the African coast Thornton (1992:32)1444 The Portuguese start extracting gold and slaves

from SenegalThornton (1992:32)

1550’s Reports of Africans traveling to France and learn-ing French

(Kerr 1812 as cited inWilson (1999)

1606 Reports of Africans speaking French "like natives" Thornton (1992:215-216)

1628 Unsuccessful attempt by the French to settle per-manently in Senegal

Biondi (1987:37)

1635 “The population of Rufisque2 speak a fairly intel-ligible kind of French [...] they pronounce in ourlanguage offences and swearwords”

(Delafosse 1931:11)

1658 St-Louis-du-Sénégal is established Biondi (1987:37)1817 Slave trade from St-Louis stopped, the export

shifted to gumWilson (1999:10)

1820 First African soldiers in the French colonial army Echenberg (1986:311)1823 First all-African company in the French colonial

army((Echenberg1986:312)

1848 Slavery is abolished in France and her colonies1857 Official creation of Tirailleur Sénégalais Echenberg (1986:315)1910 Charles Mangin propagates for the use of African

troops in a potential European warMangin (1910)

1914 WW1 starts1916 The anonymous manual of FT is published Anon (1916)1916 WW1 ends1939 WW2 starts1944 24-35 West African soldiers are killed by French

officers in Dakar, Senegal, following a mutiny3Echenberg (1978)

1945 WW2 ends1960 All states of West Africa are independent Chafer (2002)

Table 1: Brief historical overview of important events in the history of Les TirailleurSénégalais

2Rufisque is a coastal town in Senegal. It was an important town in its own right, but has now becomea suburb of Dakar, the capital. In Wolof the town is called Tëngéej.

3The movie Camp de Thiaroye by Ousmane Sembène depicts this story.

10

3.2.2 The pidgin of Français Tirailleur

The language of the West African soldiers in the French colonial army has been men-tioned in descriptive works from the 19th century and forward. The earliest documentedutterances in FT are found in Dupratz (1864).

(2) Français Tirailleur (FrTir1234, Dupratz 1864:398)Original Mission n’y a pas bon; toujours en la classe, toujours en l’étude; nonmission n’y a pas bon

Stdzdgloss

Missionmission

nNEG1

yaya

pasNEG2

bon;good;

toujoursalways/still

enin

laDEF

classe,class,

toujoursalways/still

enin

l’-étude;DEF-class

nonno

missionmission

nNEG1

yaya

pasNEG2

bongood

‘The mission is no good. There’s too much studying. No the mission sure is nogood’

Maurice Delafosse wrote about FT in 1904, describing it as a French equivalent to themore well-known English pidgins of the area.

The most cited source on the language variety is an anonymous manual, Le françaistel que le parlent nos tirailleurs Sénégalais (see figure 2). The manual was printed in 1916and was intended to facilitate the communication between French officers and the Africansoldiers in the French army.

Figure 2: Cover of the anonymous manual of FT. c�1916 Imprimerie Militaire Universelle

11

The manual is prescriptive, informing white officers how they should best formulateorders for optimal effect. The author(s) does make comments that suggest that the materialis based on some actual experience with West African soldiers. There is also references tothe structure of Bambara that implies that the FT found the anonymous manual is theproduct of a conscious effort rather than natural utterances.

The manual is divided into two parts, a description of the language variety (here labeled‘DESC’) and a set of commands and phrases (‘COMM’). Wilson (1999) has analyzed themanual extensively and she suggests that it is possible that the two parts were written bytwo different people. The material in the anonymous manual concern the life of the soldierand war, there is little mention of anything not related to the military. (3) is an typicalexample of a sentence from the anonymous manual:

(3) Français Tirailleur (FrTir0257a, Anon 1916)OriginalSi ennemi ya gagné blessés trop, tués trop

Stdzdgloss

Siif

ennemienemy

yaya

gagnerhave/get

blesserwound/wounded

trop,much,

tuerkill/killed

tropmuch

‘If the enemy has many wounded and killed’

Delafosse (1904) and the anonymous manual of 1916 do not only contain utterancesin FT but also statements about the grammar of FT. The manual has been analyzed byWilson (1999) and she provides further insights into FT based on the manual. There isalso a brief description of FT in Corne (1999). The next section of this thesis contains asummary of the statements that have been made about the linguistic structure of FT.

3.2.3 Summary of previous statements about the grammar of FT

This section contains a list summarizing the statements that have been made about thestructure of FT Delafosse (1904), Anon (1916) and Wilson (1999). The statements fromAnon (1916) are also accompanied by page reference to the thesis by Wilson as it containtranslations into English and a valuable commentary.

Anon (1916) often gives para-linguistic advice to French officers on how they shouldcommunicate with the black soldiers, for example Désigner toujours le même objet ouexprimer la même idée par le même mot [translation into English by Wilson: always usethe same word to express same object or idea]. Statements about FT in this manner are notincluded in here, this summary contains explicit statements about the linguistic structureof FT.

I have aimed at summarizing the statements in a clear and concise manner. As said be-fore, there is variation in FT and there are statements in this summary that will contradicteach other. I will first give a summary of statements on FT made by previous literaturethat concerns this investigation directly. Following that there is a summary of statements

12

that have been made about FT in the literature, but that will not be discussed directlyin this study. The list of statements is structured in the same way as the result section ofthis thesis.

Polar interrogation. Anon (1916:16) and Wilson (1999:47) claim that polar interro-gation in FT is expressed by intonation alone. This is addressed in section 5.2.1 of thisthesis.

Standard/verbal Negation (SN). Wilson (1999:92) claims that SN is expressed bypre-verbal pas alone, but follows, what she labels, the finiteness marker and relativizers yaand yena. (Delafosse 1904:265) claims that pas is positioned after the verb. Section 5.2.2of this thesis is devoted to the expression of SN.

Grammatical gender. Anon (1916:7) claim that all inanimate objects are assignedthe etymologically masculine gender (le, un, mon). This can be interpreted as statementof FT lacking a distinction in grammatical gender. Furthermore, the anonymous manualclaims that the natural gender of animate nouns is expressed by the suffix femme, i.e.chien-femme means ’bitch’. Section 5.2.4 of this thesis is devoted to the expression ofgender in FT.

Anon (1916:7) and Delafosse (1904:265) says that there are no articles (indefinite ordefinite). There are however cases where the indefinite article is perceived as belongingto the word, for example: mon latête. The issue of agglutinating articles and possessivepronouns is also dealt with in section 5.2.4.

Personal pronouns. Anon (1916) claims that there is no distinction between subjectand object in the personal pronoun system. FT doesn’t have any special forms for reflexivepronouns either (Wilson 1999:92). The pronouns are: moi, toi, lui, nous, vous and eux(Anon 1916:11). An overview of the personal pronouns of the FT-corpus is found in 5.2.5.

Attributive possession. Anon (1916:9), Wilson (1999:34) and Delafosse (1904:265)gives several strategies in FT for the expression of attributive possession (constructionswhere the possessum7 and the possessor form a noun phrase).

When the possessor is a 1st or 2nd person singular pronoun FT is said to employpossessive pronouns (mon, ton). For the other persons and nominal possessors FT has aprepositional construction with pour ‘for’. In the pour -construction the order is possessum- pour - possessor. Delafosse (1904:265) also describes a prepositional construction, butwith de ‘of’ instead of pour ‘for.

Attributive possession can also be expressed through juxtaposition (no prepositions, nopossessive pronouns). The anonymous manual says that the preferred order is possessor- possessum, but the opposite is possible as well. Example: tirailleur fusil, meaning ’thesoldier’s rifle’ [sic] (Anon 1916:14 and Wilson 1999:43, 84).

It is noted by the anonymous author(s) that the order possessor - possessum is mostlikely a result of influence from substrate languages. If a speaker uses possessum - pos-sessor, the anonymous manual (page 15) says that “ils seront obligés de faire mentalementl’inversion pour rétablir l’ordre qui, pour leur langue, est l’ ordre naturel” [Eng: ‘they willhave to mentally reverse the order so as to retain what is for them the natural order’]. In

7Possessum: that which is possessed, also known as ’possessee’.

13

other words, to facilitate communication with the soldiers the officers are encouraged touse structures the author(s) believe are present in the substrate languages.

The expression of attributive possession is found in section 5.2.6 of this thesis.Predicative possession. Anon (1916:13) and Wilson (1999:44) states that the most

frequent expression of predicative possession is a construction involving ya and gagné.Predicative possession is discussed in section 5.2.7 of this thesis.

Copula relations and adjectives. Copula is a linguistic term that can have manydifferent meanings in the literature. Wilson (1999:81) Corne (1999:2000) says that FT hasno copula, and I believe they are referring to and equative, predicative and/or identity-copula. The anonymous manual notes that the French copula-verb être ‘be’ doesn’t exist inFT and that it is sometimes replaces by ya (Anon 1916:13). Wilson writes that stative verbsya and yena can be used in similar contexts as predicative copulas8 This has implicationfor the interpretation of adjectives as a part-of-speech in FT. Copula relations are foundin section 5.2.8 and adjectives are discussed in section 6.

Anon (1916:8-9) claims that adjectives in FT are not expressed by a modifying wordwithin the noun phrase but rather with a relative clause containing a ya or yena and a 3rdsingular pronoun. Anon illustrates this with the example that un enfant bon [Eng: a goodchild] becomes un enfant (que) il est bon [Eng: a child (that) he is good] (Anon 1916:8-9and Wilson 1999:33). It is not clear if these are to be seen as finite clauses or not.

Wilson (1999:81) states, after analyzing the phrases listed by the anonymous manual,that there are adjectives expressed in the same manner as in the lexifier and that they canbe pre-posed an post-posed what the modify.

Moyen. (Wilson 1999:52) notes that the notion of ‘be able to’ is moyen (as opposedto French pouvoir ‘be able to’), this is investigated in section 5.2.9.

Ya . One of the most frequent items in the corpus and one that has been associatedwith FT more than any other is ya. Wilson (1999:81, 87) analyzes ya as a stative verb(when occurring alone) and as a marker of finiteness (when in combination with otherverbs). It is not clear what finiteness is defined as since there is no overt marking of tense,aspect or mood.

Furthermore, ya is not a copula and all finite verbs are preceded by ya. The stativeverb-ya can denote location/existence (’there is/there are’) (Wilson 1999:88-89). Wilsonalso notes that there are instances where ya marks predictive possession. The distributionand potential function of ya is discussed in section 5.3.2.

Yena. This marker has also had a lot of attention. Wilson (1999:81) describes it asa stative verb (when alone) and a relativizer. Anon (1916:13) notes that it can replaceêtre ‘be’, but that it is more common in relative classes that are preceded by qui/que inFrench. Delafosse (1904:265) notes both ya and yena as particles of the verb, no furtherdetails. Yena is discussed in section 5.3.3.

Reduplication. (Corne 1999:201) writes that reduplication is a frequent strategy formarking intensity or continuity, using (4) as an example. Reduplication is discussed in thisthesis in section 5.3.6.

8Why this does not then qualify ya and/or yena as copula-verbs is not clear.

14

(4) Français Tirailleur (FrTir0328, Anon 1916Original tirailleur ya besoin tirer, tirer, tirer toujours

StdzdGloss

tirailleurtirailleur

yaya

besoinneed

tirer,shoot

tirer,shoot

tirershoot

toujoursalways/still

‘The tirailleur should shoot without stopping’

Gagner. (Delafosse 1904:265) suggests that gagner ‘win/acquire/get’ can mean ‘be-come’ and mark past tense, but it is not clear how strong his claim is. (Wilson 1999:91)notes that ya gagner/gagné can form a passive construction. Section 5.3.4. of this thesiscontains details on gagner in the corpus.

Content. (Wilson 1999:53) writes that ya content means ‘want to’ (as opposed toFrench vouloir ‘want’), this is dealt with in section 5.3.5.

Statements not investigated directly in this thesis

Verbalizer. New verbs can be created using the verbalizer, faire ‘do’ (Wilson 1999:86 andCorne 1999:200). The construction faire manière + a verbal element means ‘to try’. Incombination with a nominal element faire manière means ‘to use’. If it precedes a claus itmeans ‘to act in such a way as to’ Wilson (1999:86-87).

The construction faire mêmechose is a periphrastic construction that means ’to actlike something’. It is a strategy used to overcome gaps in vocabulary (Wilson 1999:87 andCorne 1999:200).

Demonstratives. Demonstratives pronouns are expressed by ça ’that/this’ or .. yena là (roughly ’that which is there’). Example: ça tirailleur or tirailleur yena là (Anon1916:9 and Wilson 1999:34). Delafosse (1904:265) also describes the use of là ’there’ asdemonstrative.

Prepositions. There are few, if any, prepositions (Anon 1916:14 and Wilson 1999:43.The preposition à (and therefore also the portmanteaus au and aux ) is deleted (Delafosse1904:265). Example: je vais au village becomes moi parti village.

Coordination is asyndetic. Conjunctions are often, if not always, omitted (Wilson1999:93-94).

Nominal morphology There is no singular/plural distinction on nouns (Anon 1916:8and Wilson 1999:31).

Verb morphology. The main verb is always in what is described as “the simplestform”. This often what is in the lexifier French called the infinitive form (Anon 1916:12and Wilson 1999:39), but (Delafosse 1904:265) notes that it can also be the past participleor imperative form of French. This means that there is no inflection on the verb for person,gender or number.

Mood. Wilson (1999:106) writes that there is no formal distinction of mood. Impera-tive clauses are however often subject-less (Wilson 1999:91 and Corne 1999:201).

15

Aspect Wilson (1999:106) writes that there is no marking of aspect. (Delafosse1904:265) describes a verbal construction that could be interpreted as perfective past orpresent perfect. It is formed with a/ ya gagné + past participle. Example: lui ya gagnémort meaning ’he is dead’.

3.2.4 The linguistic makeup of the Français Tirailleur troops

There are many languages that have been mentioned in connection with FT. First of allthere is great consensus on the matter of the lexifier (the language that has contributedwith most lexical items), it is French (as spoken in 1800’s) . The primary contact with theFrench language must have been through spoken discourse, which means that it is unlikelythat structures and words that were uncommon in the spoken language at that time couldhave made it into the pidgin.

There was a French officer by the name of Charles Mangin in the late 19th century andbeginning of the 20th century. In 1910 he published a book called La Force Noire in whichhe propagates for the use of African troops in the event of a European war. This book alsocontains valuable information about the African troops and their composition.

Figure 3: Cover of La Force Noir by Mangin (1910). c�1910 Charles Mangin.

Mangin (1910:274-275) writes that the first troops of African soldiers were mainlycomposed by the Wolofs and Toucouleurs. The Toucouleurs 9 are a part of the Fulani group.

9The term ’Toucouleur’ is French popular etymology based on the fact that these people are believed

16

Both the Wolofs and the Fulani speak languages of the North-Atlantic branch of the Niger-Congo family and are predominately muslim. The Fulani group is a large ethnic groupin West Africa that has a wide geographical spread and many different dialects/languages(Harrison 2003). We cannot be sure what variety of the Fula was spoken by these WestAfrican soldiers.

Biondi (1987:49) points out that one of the biggest differences between slavery in WestAfrica and the new world (the americas) was the presence and importance of the “mixed”population and ’signares’ in particular. ’Signares’ is a term used for African or part-Africanwomen who were companions to the French men of the colony.

Thus, the first troops were made up of speakers of Wolof and Fula. It is not unlikelythat there were soldiers, speakers of Fula in particular, who had some knowledge of Frenchprior to enrollment.

After Wolof and Fula, the Serer (also Northern Atlantic; Niger-Congo) were added tothe troops. At the time of publication of La Force Noir the most dominating group wasMande (Bambara, Mandinka, Mende, Dyula, Soninke and Susu) and they were recruitedafter the Serer. The languages spoken by the Mande group are not related to the Niger-Congo. The Bambara and Mandinka were recruited first and later Susu and Dyula10 andlastly Soninke.

The last group that Mangin notes that the French military recruited were the Hausa ofDahomey (modern day Northern Benin). The Wolof and Toucoulers (Fula) were preferredin the beginning according to Mangin because they were easier to incorporate in the militaryas there were already indigenous officers who spoke their languages (Wolof and Fula). TheBambara proved more difficult to instruct at first since they did not speak Wolof nor Fula,but they did later make up the largest group of the West African troops in the Frenchcolonial army.

The Hausa language is not related to Niger-Congo or the Mande languages, but amember of the Afro-Asiatic language family. This means that there were languages fromat least three separate language families spoken among the soldiers: Niger-Congo (NorthernAtlantic), Mande and Afro-Asiatic.

Van Den Avenne (2012:258) underlines the influence of Bambara in FT; she describesthe anonymous manual from 1916 as portraying FT as a “calque11 de la langue Bambara”.

Table 2 contains information on the different languages that have been mentioned inconnection with FT, primarily by Mangin. The genealogy is taken from WALS and theinformation on geography from Ethnologue. We cannot be sure that all of these languagesare indeed involved in FT, but they have been proposed as such.

to originate from the state of ’Takrur/Tekrur/Tekrour’, a West African state in 800 - 1285. The Frenchhave reinterpreted that name as meaning ‘Toucouleur’ - all colors.

10“Dyula” is not to be confused with Dioula-Fogny, another Niger-Congo language, but of the NorthernAtlantic branch.

11Calque is a linguistic term, mainly used in French literature. It denotes a word-by-word translation-

loan where each lexical element is replaced (in this case by a French word) but the structure is the same.In this context it is possible to say that she describes the FT of the manual as relexified Bambara.

17

Language Family & Genus Geographical spreadFrench Indo-European, Romance France & West AfricaWolof Niger-Congo, Northern Atlantic Senegal, Gambia, Maurita-

niaSerer/Noon Niger-Congo, Northern Atlantic SenegalFula (Senegal) Niger-Congo, Northern Atlantic Senegal, Mali, Sierra Lione,

GuineaMòoré Niger-Congo, Northern Atlantic Burkina Faso, Mali, TogoDyula Niger-Congo, Western Mande Mali, Côte d’Ivoire, Burk-

ina FasoBambara Niger-Congo, Western Mande Mali, Côte d’IvoireMandinka (Gambian) Niger-Congo, Western Mande Senegal, Gambia, Guinea-

BissauMandinka (Senegal) Niger-Congo, Western Mande Senegal, Gambia, Guinea-

BissauManinkakan (Western) Niger-Congo, Western Mande Senegal, Gambia, MaloMende Niger-Congo, Western Mande Sierra Lione, LiberiaSusu Niger-Congo, Western Mande Guinea, Sierra LioneSoninke Niger-Congo, Western Mande Mali, Gambia, Guinea-

Bissau, Mauritania, SenegalHausa Afro-Asiatic, Western Chadic Nigeria, Benin, Burkina

Faso, Cameroon, Ghana,Niger, Sudan

Table 2: Languages proposed as involved in the multilingual situation of FT.

3.2.5 Summary: Les Tirailleurs Sénégalais, West Africa and France

West Africa became known to European powers in the 14th century. France had a greatinfluence over the region during the 17th and 18th hundreds, after the ‘Scramble for Africa’France was in possession of great territories inland as well.

The French colonial army enrolled native people of Africa, the soldiers of West Africawere called Les Tirailleur Sénégalais. The West African troops were made up of people ofmany different ethnic origins, they did not share a common language.

There is mention of a contact language variety derived from French in these troopsand by 1916 the French military even publishes a manual of the pidgin. Most previousliterature have referred to this manual. It contains much information, both explicitly andimplicitly. There is also the work of Maurice Delafosse describing FT.

Delafosse (1904), Wilson (1999), Corne (1999) and the anonymous manual describe areduced language variety which lacks morphology and differs from the lexifier French in

11RDC = République Démocratique du Congo

18

certain respects. The anonymous manual also notes that there is substrate influence, pri-marily from Bambara. FT is said to lack prepositions, conjunctions and question particle.There are two forms that are very frequent and characteristics: ya and yena. The literatureis not clear on the nature of these forms; they are described as stative verbs, markers offiniteness and relativizers.

FT is said to lack overt copula and have different strategies of filling lexical gaps, suchas periphrastic constructions with verbalizers.

There are many West African languages that are likely to have been present in themultilingual situation that gave rise to FT, it would seem, however, that we have reasonto believe that Bambara, Fula and Wolof might be particularly influential.

19

4 Method

This is a corpus study of manually annotated material. The corpus was collected, trans-lated, standardized and glossed by Mikael Parkvall. The annotation has been done by me.Mikael Parkvall is a well-known researcher and has published many articles and books oncontact languages, among others his PhD dissertation on substrate influences on AtlanticCreoles (Parkvall 2000).

Each string is accompanied by meta-data: bibliographical information and notes ondate of production. If the original source contains any information about the speaker(ethnicity or gender) this has also been included.

One important issue is the spoken nature of FT. Pidgins are less often put into printand many of the utterances found in the corpus are representations of speech as opposedto written text that was intended to be read. In a spoken discourse, it is possible for thespeaker to make use of situation specific information, shared knowledge or gestures. Allthis extra information is lost in writing and this makes the material more difficult to workwith.

4.1 Source material

This section contains information about the documents that constitute the corpus, specialattention is devoted to three documents that have contributed greatly.

This corpus contains material from 148 different documents by 130 authors. The oldestdocument was published in 1864 (Dupratz 1864) and the most recent 2008 (Ruault 2008).The date of publication is not necessarily the same date as the author attributes to theproduction of the utterance. This means that while Dupratz still contains the oldestutterances, but the utterances with the most recent date of production are found in Biasini(1995) and Bellaigue (2009). These are attributed to the 1950’s.

Table 3 shows the strings of the corpus as they are distributed over time of utterance.Several documents contain material that is attributed to several different dates. Table 4shows the eight documents that have contributed the most to this corpus. The two sectionsDESC and COMM of the anonymous manual have been treated as two different documentsby two different authors.

20

Words Strings Authors DocumentsTotal 8 686 1 737 130 148Before 1916 4 564 945 82 102After 1916 4 099 788 43 431860-1869 67 13 12 161870-1879 197 46 5 51880-1889 564 116 12 171890-1899 2 176 448 34 401900-1909 765 160 17 191910-1916 795 160 21 241916-1919 2 354 428 14 151920-1929 791 155 7 71930-1939 447 85 9 91940-1949 428 105 11 111950-1959 79 15 2 2Date unknown 23 4 1 1

Table 3: FT material sorted by year of production of utterance

Document Words StringsAnonymous 1916 (COMM) 1505 249Cousturier (1920) 573 119Baratier (1912) 327 62Diarra (1927) 285 48Lhote (1947) 253 43Marie-Victoria (1921) 211 48Desjardins (1925) 201 38Leymaire (1898) 197 34

Table 4: The eight largest documents of the corpus

Many of the documents are anthropological in nature, reports from travels by mission-aries or military personnel in West Africa. Material written by authors that have not spentany time in the area or met any West African soldiers has been excluded.

The source with the most utterances is the anonymous manual from 1916 (see section3.2.2.), followed by Des inconnus chez moi — a biographical book by Lucie Cousturier.

Lucie Cousturier was not a part of the military, she was a painter and writer. Sheencountered West African soldiers during WW1 in the French countryside. They werestationed there before going up to the front. She took it upon herself to organize classes inFrench for them (Little 2009). It was these encounters that formed the basis of her book,Des inconnus chez moi (’The Strangers/Unknowns in My Home’). She later travelled toWest Africa and wrote a travelogue called Mes Inconnus chez eux (’My Strangers/Unknownsin Their Home’), it does not however contain utterances of FT.

21

Figure 4: Cover of Épopées Africaines byBaratier (1912) featuring an illustration byLucien Pouzargues. c�1912 Albert Baratierand Lucien Pouzargues.

The third largest contributing docu-ment to the corpus is Albert Baratier’sÉpopées Africaines — a collection of re-ports and stories by a French officer whotravelled in the area. The accounts of hismilitary life and travels are accompaniedby illustrations of Africans and the localwildlife by illustrator Lucien Pouzargues.

The fourth largest contributing docu-ment is a letter that was written by a WestAfrican soldier — a Tirailleur Sénégalais —by the name of Baba Diarra. He fought forFrance in the conquest of Morocco in 1912and in the first World War. During the firstWorld War he became sergeant. The let-ter is written in FT and concerns the dis-crimination of African soldiers in the Frenchcolonial army. Diarra writes that the WestAfrican soldiers are payed less and treatedbadly (see example 5).

The letter was published in the first is-sue of the journal ‘La Race Nègre’ in 1927,which was founded by the old West Africansoldier Lamine Senghor (Dewitte 1990:128).The letter finishes with the suggestion that if war breaks out again it should be fought besoldiers who have the rights of French citizenship and not natives of the colonies.

(5) Français Tirailleur (FrTir0792, Diarra 1927)Original Gradé ropéens y a compter nous comme sauvasi, comme plus mauvaischien encore

StdzdGloss

Gradéofficer

européenEuropean

yaya

comptercount

nous1pl.ACC

commeas

sauvage,savage,

commeas

plusmore

mauvaisbad

chiendog

encorestill/again

‘The European officers saw us as savages, even worse than dogs’

22

4.2 Reliability of the sources

A problem with this study is the reliability of the sources. Authors can be notoriouslyinconsistent and potentially untrustworthy due to prejudices or lack of insights into thelanguage. Contact language varieties are often associated with a negative and/or exoticstereotype. This can give rise to authors producing utterances that are suspiciously similarto the lexifier or, the other extreme, made more “pidginy” and exotic.

There are utterances included in the corpus that are potentially untrustworthy, it isimpossible to know for certain whether or not a author has misunderstood or consciouslymanipulated the material. Parkvall has in the collection of the material attempted toexclude utterances that are most likely not authentic FT. Two important reasons for ex-clusion has been (i) excessive glorification of France and (ii) author hasn’t spent any timein the area.

The analysis of the data in the result section is based on utterances from several differentsources (unless otherwise indicated), i.e. I want to avoid features that are present in one orvery few documents. Most of the features in the result section have not suffered from thisproblem, the documents are quite alike with respect to what is investigated in this thesis.

4.3 Comparison with other languages

When investigating the structure of FT it is interesting to compare not only with thelexifier of French but also potential substrate languages and linguistic typology.

In order to make this comparison, I have consulted linguistic reference literature andmade great use of the World Atlas of Languages Structures (Dryer and Haspelmath 2011).WALS is a large database of grammatical, lexical and phonological features in the languagesof the world. The database was constructed by an author team of 55 scholars. The sampleof languages varies with the different features. WALS aims at having a balanced sampleof the languages of the world, both genealogically and geographically. In order for thedifferent features to be comparable, there is a special set of 100 and 200 languages that areincluded in most chapters.

It is important to keep in mind that in many WALS chapters, the languages are onlycoded with one value even though the language might sport more than one strategy for thatspecific feature. An illustrative example of this is the word order-chapters (Dryer 2011g),where several languages allow for more than one order of subject, verb and object. Theword order that is perceived by reseachers of the language as dominant is the value foundin WALS, even if other types are possible. This is also true of the chapter on expressionof polar interrogatives (Dryer 2011h) and many more.

4.4 Limitations of the study – the nature of corpus

The FT-material has been annotated for highly frequent grammatical phenomena such aspredicative possession, subject and position of negation. A full account follows in the nextsection.

23

The size of a corpus, the nature of the sample and balance determines what researchquestions it can be used for. When describing low-frequency phenomena, such as thelexicon, a large corpus is needed. This is a small corpus and it is not balanced with respectto speakers, genres, etc. This means that the study must be limited to describing highlyfrequent linguistic phenomena such as the structure of simple declarative sentences, thepronoun system etc. We cannot and should not attempt at extracting more from the dataset than this.

4.5 Annotation of the data

The data has been glossed, translated and annotated by hand to the best of our abilities.The glossing and translation has been done by Mikael Parkvall. The material has beendivided into clauses and annotated for structural features has been made by me in theprogram Microsoft Excel.

Each sentence appears in the corpus in its original form and in a standardized form. Thestandardized material was created to facilitate comparison. and is primarily concerned withcoherent spelling of very similar items (for example: émpé ! unpeu), deletion of spaces inconstructions/words that are most likely one unit (y a ! ya, y en a ! yena). There aremany cases where different verbal forms in orthographic French are pronounced identically,such as the past participle and infinite form. In the standardized material the infinitiveform is favored if there is no difference in pronunciation.

This study is not only limited by the small size of the corpus but also by the abilities ofnon-native speakers to handle the material. I am not a speaker of FT, not a native speakerof French. This is one more reason why we have focused on features that are subject toless variation and ambiguity.

The corpus has been be annotated for

• independent or subordinate clause

• type of sentence (interrogative, declarative, imperative)

• stative or dynamic verb

• expression of polar interrogation

• standard negation (ne... pas, pas, point, mie, goutte)

• non-standard negation (plus ’no more/longer’ , rien ’nothing’, jamais ’never’ andpersonne ’no-one’)

• expression of grammatical gender

• expression of pronouns

• lexical or pronominal subject

24

• drop of overt subject

• copula relations (predicative, presentative, locative, existential and equative)

• presence of certain potentially interesting specific items (ya, yena, moyen, manière,etc)

• reduplication

• potential TA-marking

25

5 Results and analysis

This section contains the results and the analysis. It is further divided into four parts.The first part (5.1) concerns the basic makeup of the data; type-token-ratio, indepen-dent/subordinate and declarative/interrogative/imperative. The second part (5.2) is la-beled “function seeking form” and contains the findings from the investigation of certainbasic functions that we expect the pidgin to express. The third part (5.3) is labeled “formseeking function” and is the mirror image of the previous. Instead of investigating the for-mal expression of an expected function I have attempted at finding a function of a certainhighly frequent forms. The fourth and last section (5.4). contains notes on change overtime in the material.

5.1 Data makeup

This section covers type/token-counts and basic categories such as independent clause/subordinateclause and stative/dynamic.

As stated previously, the difference between the original and standardized material isprimarily more coherent spelling and the deletion of spaces.

5.1.1 Clause-types and stative/dynamic

This section mainly serves as an introduction to the corpus material, one cannot drawconclusions about the structure of FT directly from these counts.

Table 5 shows the amount of independent and subordinate clauses. There are stringsthat do not consists of clauses but of unique words, these are included in the study andlabeled ‘other’ here.

Independent clauses Subordinate clauses other1 521 (88%) 208 (12%) 8 (0.04%)

Table 5: Independent/subordinate clauses

Table 6 contains counts of declarative, interrogative and imperative clauses among thestrings. There are strings in the corpus that cannot be categorized in an obvious way, thesehave not been forced into a category, but marked as ‘other’.

Total declarative interrogative imperative other1 737 (100%) 1 602 (92%) 62 (4%) 83 (5%) 90 (0.05%)

Table 6: Declarative/interrogative/imperative/other

26

Table 7 displays the distribution of clauses into the types ‘dynamic’ and ‘stative’. Thestative clauses are further divided into 5 different copula relations (see section 5.3.8) and1group of ‘other statives’. Other very frequent statives are ‘know’, ‘be able’ and ‘like’.

Total Dynamic All stative All COP1 737 (100%) 823 (47%) 908 (52%) 498 (29%)

Table 7: Stative and dynamic clauses

5.1.2 Type-token-ratio in the corpus

A token is one occurrence of one word. A word is here defined as a number of characterssurrounded by spaces12. A type is the the total occurrences of a word that is spelled thesame 13. The data is not lemmatized, types are defined by their orthographical expression.

The original material contains 10 083 tokens and 1 464 types. If we exclude hapaxlegomena14 there are 705 types. The standardized material consist of 8 696 tokens and 1172 types (629 without hapax). This material has been divided into 1 737 strings. Thereis material included in the corpus which does not consist of full clauses, there are 62 suchstrings.

Pidgins are said to be characterized by a restricted vocabulary (David 1971:15 andHolm 1988:73). Anon (1916) even writes that it is important to avoid synonyms whencommunicating with the West African soldiers.

“Mais il faut également [. . . ] réduire le plus possible le nombre des motsemployés et, par conséquent, éviter d’exprimer la même idée par plusieursmots différents, ce qui dérouterait l’indigene et lui rendrait la compréhensionde notre langue très difficile.” (Anon 1916:17)

Translation into English by Wilson (1999:48)But equally necessary are [...] to reduce as much as possible the number of wordsused and thus avoid using several different words to express the same idea, asthis will confuse the native and make it hard for him to understand our mostdifficult language.

If speakers of FT use less synonyms and have a restricted lexicon we would expect thisto show in the corpus material. One way to measure the lexical diversity in a corpus isthrough the type-token-ratio (TTR). TTR is the number of types divided by the number oftokens. A low TTR would suggest that a few types are being used very often. This could

12The amount of tokens will differ between the original material and the standardized since part of thestandardization involved the deletion of spaces.

13This means that she loves you, yeah, yeah, yeah for example contains 6 tokens but only 4 types.14Hapax Legomena are types with only one occurrence.

27

be due to a small corpus, limited lexicon or little/no morphology15. This measurementtends to be even lower for spoken corpora.

In order to compare small and large corpora fairly one can also use a measurement ofstandardized TTR (sTTR). sTTR is calculated in the same way as TTR, but of 2 00016

concatenated words of the corpus (Baker 2006:52).Table 8 displays the amount of types, tokens and type-token-ratios of the FT-corpus.

The three most frequent types have also been included. ‘Type (�2)’ refers to the numberof types that have 2 or more occurrences, i.e. all types minus hapax legomena.

Token Type Type (�2) TTR sTTR Freq 1 Freq 2 Freq 3Original 10 082 1 464 705 14% 27% a (866) y (674) moi (355)Stdzd 8 696 1 172 629 13 % 26 % ya (682) moi (357) pas (318)Gloss 8 765 1 052 666 N/A N/A sg (940) ya (670) NEG (339)

Table 8: Types/tokens in FT

A sTTR of 27% is quite low, but perhaps not as low as one might have expected. Wecannot compare sTTR directly across languages, but here is brief comparison with Englishto give the reader some point of reference. English is not an isolating language (Bickel andNichols 2011) and the sTTR of of the spoken material of the British National Corpus is33% Baker (2006:52). We need more data to compare corpus’s properly, and ideally wewould like to compare FT with spoken French of the 19th century.

FT, like many (if not most) pidgins, seem to lack morphology, which would pull downthe sTTR. The material in this corpus is most likely primarily spoken and it is small. Allof these factors should indicate a low sTTR, but perhaps 27% isn’t as low as we expected.This is however very speculative and we need more detailed data on the distribution oftokens and types in spoken informal French and other pidgins to further this issue.

Another way of investigation types and tokens in corpora is by the application of Zipf’sLaw. Zipf (1935) found that many natural phenomena follow a pattern where the secondmost frequent item is half as frequent as the most frequent, the third half as frequent asthe second most frequent, and so on.

This formula is known as Zipf’s Law and it is predicted to be true of such differentthings as size of population in cities and word frequencies in language. In language, it seemsto work quite well (often with the exception that the first 100 items are often irregular).

Many corpora of natural language follow this pattern, but does it also apply to pidginsin general and FT in particular?

Figure 5, 6 and 7 illustrate the type frequency of the material17. The x-axis is the types,and the y-axis the frequency of those types. The types are sorted after frequency on the

15Provided we are defining types by their orthographical/formal expression and not by their lemma-form.16Some authors use 1 000, the default value of the latest version of WordSmith. 2 000 is used in this

study.17Keep in mind that there are fewer types in the standardized material as opposed to the original.

28

x-axis. The figures display the original source material (red), the standardized material(blue) as well as what the perfect Zipf’s curve would be (black).

Figure 5 displays the frequency of all types, figure 6 of only the first 100 and figure 7is all types logarithmically. All frequencies are normalized and both standardized andoriginal material are significantly similar to the perfect Zipfs (p<0.05).

Figure 5: Type-frequency-ratio of the FT-corpus. X-axis = types ranked by freq, Y-axis=frequency. Non-Logarithmic.

29

Figure 6: Type-frequency-ratio of the FT-corpus. Non-logarithmic, zoomed in at first 100types.

Figure 7: Type-frequency of the FT-corpus. X-axis = types ranked by freq, Y-axis=frequency (log-log=10)

30

5.1.3 Summary: data makeup

There are more independent clauses than subordinate, more declarative than interrogativeor imperative. We cannot derive structure of FT directly from these numbers, but we candeduce that we have more data for describing declarative independent clauses than anyother type of clause.

An analysis of TTR of the material reveals that further investigation into the lexiconof pidgins is needed. Despite the anonymous manuals advice to avoid synonyms and theoverall characteristics of pidgins to have a restricted lexicon, the TTR is perhaps not aslow as could have been expected.

Zipf’s Law applies to this corpus, even though it is a small corpus of a reduced variety.

5.2 Function seeking form

5.2.1 Polar Interrogative

There are two chapters in WALS (92 & 116) devoted to the expression of polar interrog-atives, both written by Matthew Dryer. Chapter 116 concerns the expression of polarinterrogatives and 92 covers the position of the question particle (if there is one).

The most common strategy found in these languages is the question particle (61%,584/954 languages). 130 of these languages have the question particle in initial positionand 313 in final. The second most common strategy is interrogative intonation only (18%,174/954). In Dryer’s study, a language is only coded as having ‘intonation only’ if that isindeed the only means of expressing polar interrogation. This suggest that many (if notmost) languages make use of applying question intonation18 to declaratives in order to forma polar interrogative even if they have other strategies as well. Dryer notes one exceptionto this, there is not interrogative intonation in Imbabura Quechua. He also comments thatit is very likely that the type ‘intonation only’ is underrepresented in his survey.

The third most common strategy is interrogative verbal morphology (17%, 164/954).There is only one language in the survey that reportedly has no distinction between declar-atives and interrogatives, and this is Mixtec (Chalcatongo) of the Oto-Manguean family.

It is interesting to note that the strategy used by several Germanic languages, butalso Spanish and French, — change of word order (7) — is typologically quite uncommon.Dryer found that 13 languages (1%) of his balanced sample of 954 used this strategy.

French French has three main strategies for expressing polar interrogatives.18Interrogative intonation does not necessarily mean rising intonation.

31

(6) French, interrogative intonation (own example)

tu2sg

veuxwant.PRS

mangereat.INF

ça?DEF

‘Do you want to eat that?’(7) French, change of word order (own example)

veuxwant.PRS

tu2sg

mangereat.INF

ça?DEF

‘Do you want to eat that?’(8) French, intial question particle: (own example)

EstCOP

ceINDEF

queREL

tu2sg

veuxwant.PRS

mangereat.INF

ça?DEF

‘Do you want to eat that?’[Lit: ’Is it so that you want to eat that?’]

Dryer (2011h) has coded French as having a sentence initial particle since the interrog-ative intonation is not the only strategy and the change of word order is very infrequent(Harris 1988).

Several of the languages the might be involved in the origin of FT are found in Dryer’schapters on polar interrogatives in WALS. Table 9 presents an overview of the expressionof polar interrogation in certain West African that potentially involved in the multilin-gual situation of FT. All data is from Dryer (2011h) and Dryer (2011j) unless otherwiseindicated.

intonation only Baoule, Diola-Fogny, Koyra Chiini (Songhay), Lin-gala and Mende

initial question particle Woloffinal question particle Bambara, Fula (Cameroon), Hausa, Maninkakan

(Western), Mòoré, Serer/Noon, Soninke andFongbe (Lefebvre and Brousseau 2002)

question particle in either position Koyraboro Senni (Songhay) and Malagasy

Table 9: Polar interrogatives in certain West African languages

Most of the West African languages in table 9 employ a particle, primarily in a finalposition.

Out of the 1737 clauses in the corpus of FT, 43 are polar interrogatives. 37 of those aremost likely examples of ‘intonation only’. They appear to be identical to the declarativesentences with the exception of a question mark — leading us to believe that there isinterrogative intonation. As we have no audio recordings of FT we cannot be certain.There is no significant change over time nor is there any significant variation across authors.

32

There is however also the issue of subject drop (9), 18% of the FT clauses lack overtsubject marking. This means since change of word order in French involves the overtexpression of the subject, it is impossible to distinguish between ‘intonation only’ and‘change of word order’. There are 7 instances of subject drop in the polar interrogatives.

(9) Français Tirailleur (FrTir0051, Cousturier 1920:35)Original Y a pas content avec nous?

StdzdGloss

Yaya

pasNEG

contentbe.happy/like

avecwith

nous?1pl

‘Don’t you like us?’

While there is no way of knowing where this dropped subject is positioned I believethat the most appropriate way to treat these cases are that they display interrogativeintonation. There is a lot of subject drop in non-interrogative clauses as well, making theonly formal distinction the presence of a question mark.

There are two clear examples of word order change in expressing polar interrogatives.The first is said to have been uttered by a native speaker of Wolof (10) and the second bya white native speaker of French (11).

(10) Français Tirailleur (FrTir0845, Sonolet 1911:44)Original Dis, mon colonel, veux-tu du cadeau de Patey-Sar?

StdzdGloss

Dire,say

monmy

colonel,colonel

veux-tuwant-you

duART

cadeaugift

deof

PateySar?Pateysar?

‘Say, colonel, would you like a present from me?’(11) Français Tirailleur (FrTir0041, Cousturier 1920:40)

Original Y a toi content pour lancer grenade?

Stdzdgloss

Yaya

toi2sg.NOM

contentbe.happy/want

pourfor

lancerthrow

grenade?grenade

‘Do you enjoy throwing grenades?’

These two instances of word-order change are most likely very rare ways of markingpolar interrogation in FT and the fact that one was uttered by a native speaker of Frenchmakes it very suspicious indeed.

These is no example of the lexifier particle est-ce que. Question particles are typolog-ically common and present in most of the potential substrate languages, including pre-sumably much influential Wolof, Bambara, Fula and Maninkan. However, the questionparticle of French is sentence initial. Only three of the potential substrate languages of FTin Dryer’s sample employed initial particle.

33

It would seem that FT employs ‘intonation only’ to express polar interrogation, but wecannot be certain. Anon (1916:16) and Wilson (1999:47) writes that FT uses ‘intonationonly’, but in order to be certain we need to consult recordings (most likely impossible).

We do not know how many language employ intonation when expressing polar inter-rogation, we only know how many use intonation as the only strategy. Perhaps furtherlight can be shed on the use of intonation in pidgins when The Atlas of Pidgin and cre-ole Language Structures is released, it will feature a chapter on the expression of polarinterrogation.

5.2.2 Standard Negation

The most common strategy for expressing Standard Negation in (SN) in FT is with the pre-verbal particle pas (12). Standard negation is the negation of an entire verbal declarativemain clause (Miestamo 2007).

(12) Français Tirailleur (FrTir1182, Barret 1888:324)Original Moi pas mirer lui

StdzdGloss

Moi1sg.NOM

pasNEG

mirerlook

lui3sg.ACC

‘I can’t see him’

If we look at the languages of the world, we find that particles are the most commonstrategy. In the typological survey by Dryer (2011c) 43% (504/1159 languages) employed aparticle. The second most common strategy is a negative affix on the verb, 34% (396/1159),and 10% (120/1159) had some type of Double Negation (DN). The most common positionof the negative element in this survey of the world’s languages is pre-verbal, 40% (524/1326)(Dryer 2011e).

French has two strategies for expressing SN, double negation construction of ne... pasand post-verbal pas alone. This double negation construction is older, ne is the formerstandard negator andpas is etymologically a word meaning ‘step’. Pas was at first used toreinforce the negation as a minimizer (‘I won’t walk, not a step’). In the 16th century it wasalso possible to use point (‘dot’), mie (‘crumb’) and goutte (‘drop’) in this construction,but pas was most frequent.

In modern spoken French, it is possible to express SN with post-verbal pas only (Hopper1991:26-27), this diachronic change is known as ‘Meillet’s Spiral’ or ‘Jespersen’s Cycle’.

The particle pas is also the negator of non-verbal clauses such as possession and copularelations. The particle occurs after the finite verb, but before the main verb and beforethe object (if it is not a pronominal object). The typical order can be described as:

subject-(ne)-(pronominal object)-finite verb-pas-(main verb)-(nominal object).

If there are no auxiliaries the main verb is the finite verb.

34

The order of the optional double negation construction in French (S(Neg)VNegO) isnot the most common order of double negation in SVO languages (Dryer 2011d). It ismore common that the second negator follows the object (SNegVONeg).

If we look at some of the West African languages that are potentially interesting forthe development of FT in table 10, a majority of them employ a particle.

Pre-verbal Negative particle Mende, Hausa, Soninke,Koyraboro Senni (Songhay),Malagasy

Post-verbal Negative particle Lingala, Gba ya KaraNegative affix Diola-Fogny, Fula (Cameroon),

Serer/Noon, Fula (Nigeria)Pre-verbal Negative word, unclear if verb or parti-cle

Koyra Chiini (Songhay), Wolof,Fongbe

Obligatory Double Negation Baoulé, Mòoré, Kikongo

Table 10: Expression of Standard Negation based on Dryer (2011c,i,e).

Wolof has more than one construction, a pre-verbal negative word and a negative suffix.The obligatory double negation of Baoulé is, interestingly, formed with negative tone anda post-verbal negative particle.

The most common way of expressing SN in the FT clauses is by means of the particle pasbefore the predicate. If the clause contained ya, yena, avoir or être the negator precedesthese. There are also a few instances of the French double negation construction ne... pas.Table 11 displays the expression of SN in FT.

Negator Total freqTotal SN 324pas only’ 297n’est pas 1n’étais pas 1n’a pas 12n’ y a pas 9n’ y en a pas 2point 0goutte 0mie 0

Table 11: Expression of Standard Negation in FT

There is no evidence of point, mie or goutte — pas dominates totally. Pas occurs asthe sole negator in many different sources and from very early on. This is expected, the

35

other negators were already very infrequent in the lexifier. There is no doubt that pas isthe most frequent marker of SN.

With one exception, all of the instances of pas are directly before the predicate (13)but after the stative verbs/auxiliaries/predicate markers of ya/yena/avait/étais (if theyare present). This is different from the lexifier, where pas occurs after the finite verb, evenif it is the only verb in the sentence (14).

(13) Français Tirailleur (FrTir0182, de Mandat-Grancey 1900:134)Original Toi dire toujours les noirs pas pouvoir suivre horaire!

StdzdGloss

Toi2sg.NOM

diresay

toujoursalways/still

lesART

noirblack

pasNEG

pouvoirbe.able

suivrefollow

horaire!timetable

‘You always say that black people cannot follow a timetable!’(14) French (own example)

Tu2sg

dissay

toujoursalways

quethat

lesART

noir-sblack-PL

(ne)(NEG1)

peuventbe.able

pasNEG2

suivrefollow

lesART

horarie-stimetable-PL

‘You always say that black people cannot follow a timetable!’

There is one exception where pas occurs after the verb (15), but it is only one out of297 instances of pas.

(15) Français Tirailleur (FrTir0308b, Bouquet and Hosten 1917)Original ti manges pas charognards

StdzdGloss

tu2sg.NOM

mangeeat

pasNEG

charognardvultures

‘You don’t eat vultures’

The authors of the original documents have represented the following expressions withaccents and spaces, suggesting that they are compositional: n’a pas, n’y a pas and n’ y en apas. However, if we consider the translations it is more likely that they could be analyzedas entire words, consider (16).

36

(16) Français Tirailleur (FrTir0571, Leymaire 1898:122Original Moi n’a pas connaître combien

StdzdGloss

Moi1sg.nom

napasNEG

connaîtreknow

combien!how.many

‘I don’t know how many!’

If n’a pas did indeed consist of DN and the verb avoir it is less likely that the translationwould be in the present tense. This is the case for most of the 25 occurrences of n’a pas, n’ y a pas and n’ y en a pas which suggests that these expressions might be threedistinct particles in their own right. There are however not many instances of this, themost common expression of SN is still “pas only’ and these alternative constructions occurmostly in the older material.

Pre-verbal negation with pas only is the overwhelmingly most frequent strategy forexpression Standard Negation in FT. There are other minor strategies, but they are notnearly as frequent as pre-verbal pas.

The findings are in accordance with previous statements by Wilson (1999), but contra-dict Delafosse (1904:265) who claim that pas is post-verbal.

5.2.3 Non-standard negation (jamais, rien, personne, plus)

Non-standard negation is here defined negatively, i.e. by not being SN (see section 5.2.2).SN might in some languages be used to express the negation of possession, time, existenceetc, but it is common than languages have special strategies for this (see Veselinova 2010and Veselinova 2013). In the case of French and FT, the relevant functions are the negationof time (jamais- ‘never’ and plus - ‘no longer’); and negative indefinite pronouns (rien,personne - ‘nothing’, ’no-one’).

Negation of non-verbal clauses (copula, possession, existence, location) is expressedwith the same means as SN, pre-verbal pas.

In written French, this type of negation is expressed with a double negation constructioninvolving the old negator; ne before the finite verb and jamais, plus, rien or personne after:je ne regrette rien (’I don’t regret anything’). Jamais - ‘never’ and plus - ‘no longer’ aredefined as adverbs, rien - ‘nothing’ and personne - ‘no-one’ as negative indefinite pronouns.It is not ungrammatical to omit ne in spoken discourse (Alsenoy 2011).

As negative indefinite pronouns personne and rien can also occur in sentence-initiallyas subjects with ne directly after; personne ne me connaît ici (’No-one knows me here’)and rien ne change - ‘nothing changes’.

Table 12 displays the occurrences of non-standard negation in FT.

37

English equivalent Negator Total freqno-one personne 5never jamais 19never n ya [...] jamais 3nothing rien 17nothing pas [...] rien 2no more plus 10no more n ya [...] plus 2no more n’a plus 1

Table 12: Expression of non-standard Negation in FT

Most of the instances of these words in the FT material are very similar (if not identical)in function to the lexifier. It is however very common to omit ne, i.e. rien, plus, jamaisand personne are inherently negative. As with the expression of SN, the instances of thedouble-word-construction with n’/ne are scarce and all before 1900.

Omission of ne is possible in modern spoken French in many of these instances. Theomission of ne is most likely a reflection of the spoken French at that time rather than aprocess that took place in both FT and French independently.

Both rien and personne have non-negative function in French. Rien is etymologically apositive indefinite noun meaning ’thing’ 19 and there is still a marginal use of rien meaning’very small thing’ (cf. Eng. ’ounce’) in modern French (il restera souvent un rien desuspicion - ‘there remains an ounce of doubt still’). Personne can also mean ’person’ inmodern French.

There are also occurrences of plus in the non-negative meaning of ’more’ in the FT-material, this could potentially be a problematic ambiguity.

The expression of non-standard negation in FT is very similar, if not identical to theexpression in the lexifier French. There are however not many examples of non-standardnegation, perhaps they differ more than this material suggests.

5.2.4 Grammatical gender

Some languages divide nouns into different groups called ‘grammatical gender’ or ‘nounclasses’. The term ‘noun class’ is sometimes used when there are more than two or threeof these groups. I will use ’grammatical gender’ and ’noun class’ synonymously.

The most important feature of grammatical gender is agreement; elements such asarticles, demonstratives, numbers have to agree with the gender of the noun they modify(Corbett 2011).

French divides nouns into either masculine or feminine20. The masculine gender inFrench is overall more frequent and inanimate nouns are mostly masculine. French has

19From the Latin accusative of res meaning ’thing’20The relation between the “natural” gender of masculine and feminine and these grammatical categories

is rather bleached. For example: bras are masculine (un soutien-gorge - ‘a bra’).

38

gender agreement on articles, possessive pronouns, adjectives, adverbs and past participles.Many languages of West Africa have gender systems consisting of more than two groups

(Corbett 2011). Diola-Fogny and Fula (Guinean) are coded in WALS as having ‘five ormore’, Hausa has ‘two’ and Bambara none.

Considering that there is little (or no) inflection of adjectives, adverbs or past participlesin FT, the place to look for gender agreement is in the few instances of indefinite anddefinite articles (un/une, le/la) and possessive pronouns (mon/ma, ton/ta, son/sa). Thereis a total of 322 occurrences of these morphemes in our 1737 strings and they occur allthroughout 1860-1950.

Some authors have represented the articles as fused with the noun, such as FrTir0739émpé - ‘a little’ (Diarra 1927:54). These instances are listed as ’agglutinated’ in table 13.They are not excluded from the other counts of the articles and pronouns. Cases wherethe lexifier French agglutinates the article are not counted as agglutination here, such asl’école. Most (80%) of the explicitly agglutinated articles occur in Anon (1916) COMM.

In table 13, “right” and “wrong” refers to whether or not the article or pronoun agreesin gender with the gender of that noun in French. There is no reason to assume the FThas the same genders as French, this is primarily etymological information.

Total Masculine Feminine“Right” “Wrong” “Right “Wrong”

Total 322 (100%) 194 (60%) 32 (10%) 88 ( 27%) 8 (2%)Indefinite articles (un,une)

47 (100%) 11 (23%) 2 (4%) 2 (4%) 0 (0%)

Definite articles (le, la) 137 (100%) 58 (42%) 2(1%) 77 (56%) 0 (0%)Possessive pronouns(mon, ton, son, ma, ta,sa)

128 (100%) 85 (66%) 28 (22%) 9 (7%) 8 (6%)

Explicitly agglutinated 44 (100%) 29 (66%) 0 (0%) 15 (34%) 0 (0%)

Table 13: Gender marking in FT

There are more occurrences of words that are etymologically masculine than feminine.More words are assigned masculine gender agreement when they are etymologically femi-nine than the opposite.

All of the instances where feminine gender has been assigned to a noun that has mas-culine gender in French (“wrong agreement”) were either ma capitaine (’my captain’) orma lieutenant (’my lieutenant’). This is quite peculiar, especially if we consider that evenif there were female officers (of which there is no evidence) they are to be called ’capitain’only - the possessive pronoun is only used for male officers (Bobe 2007). These examplesmay be few, but they are found in five different sources. There sources are not believed tohave influenced each other and were all published before 1900.

Anon (1916:7) and Delafosse (1904:265) suggest that the articles can be so agglutinatedso that constructions like mon latête are possible. There are four instances of this in

39

the corpus: FrTir0730 mon la poule (Fargeas 1899:131), FrTir0227c son lamain (Anon1916), FrTir0852 cette la guerre Cousturier (1920:173) and FrTir0366 ton latête (Sembène1987:2h04). 3 out of 7 occurrences of tête do not have la, 8 of 20 for guerre and 9 of 15 formain. There is only one instance of poule.

There are seven cases of the portmanteau du (de + le) and 1 of au à + le, all withcorrect agreement according to French.

As for the statement by Anon (1916) that feminine natural gender on animate nounsis expressed by adding femme as a suffix, there is only on sentence displaying this, (17).

(17) Français Tirailleur (FrTir0550, Augouard 1905:231)Original Moi voulait cochons les hommes, pasque moi y en a cochons les femmes,et cochons les femmes tout seuls, y en a pas connaître faire pitit!

StdzdGloss

Moi1sg.NOM

voulaitwant

cochonpig

lesART

homme,man

parcequebecause

moi1sg.NOMyena

yenapig

cochonART

leswoman

femme,and

etpig

cochonART

leswoman

femmeall

toutaloneyena

seul,NEG

yenaknow

pasmake

connaîtrebabies

faire petit!

‘I want a boar, because I only have sows, and sows alone cannot have piglets’

However, this seems to be the only instance where the natural gender of an animatebeing is a relevant topic of the material, so we cannot assume that it is productive, northat it is non-productive.

Anon (1916:7) claims that all inanimate nouns are treated as having masculine gender.While it is true that masculine gender is more often applied to feminine nouns than theopposite; there are several instances of inanimate nouns with feminine gender marking;FrTir0004 la bande - ‘the cartridges’ (Conombo 1989:132).

Out of 1 737 strings there are only 303 forms that mark what is etymologically gram-matical gender, and many are given masculine gender though the noun is etymologicallyfeminine. This suggests that FT does not have a productive gender system. This is alsoproposed in the anonymous manual and explicitly stated by Wilson (1999:110). The ma-jority of the contact varieties with French as a lexifier in APiCS21 do not show genderagreement on adjectives or indefinite articles (Maurer and the APiCS Consortium 2013).

5.2.5 Personal pronouns

The French personal pronoun system is rather complicated and makes distinctions betweennominative, accusative and dative case. There are also reflexive forms and so called “dis-

21Most contact varieties in this sample are traditionally considered creoles, not pidgins or mixed lan-guages.

40

junctive” pronouns. The disjunctive pronouns are used in prepositional phrases and certainempathic contexts.

Table 14 gives a basic overview of the French pronominal system.

NOM ACC REFL DAT DISJUNCTIVE1sg je me me me moi2sg tu te te te toi3sg il, elle, on, ça, ce le, la, en se lui, y lui, elle, soi1pl nous nous nous nous nous2pl vous vous vous vous vous3pl ils, elles, les en se leur, y eux, elles

Table 14: French pronouns

Anon (1916:11), Wilson (1999:37) both claim that the personal pronoun system of FTmakes no distinction between nominative case and non-nominative case and indeed thisseems to hold up also for material other than the anonymous manual. Table 15 displaysthe expression of pronouns in FT. There is variation in the material, the table displaysthe most frequent form followed by absolute number of occurrences and percentage of alloccurrences of that function. For example: moi occurs 303 times as 1st person singularin subject position (NOM), that is 95% of the times there was a form marking 1st personsingular pronoun in subject position in the entire material. The marking of pronouns inwhat is etymologically accusative and dative case have been lumped together as there wasno difference in FT.

1sg 2sg 3sgNOM moi 303 (95%) toi 184 (89%) lui22176 (87%)ACC&DAT moi 41 (89%) toi 44 (100%) lui 43 (91%)

1pl 2pl 3plNOM nous 73 (99%) vous 9 (100%) eux 18 (45%)ACC&DAT nous 21 (100%) vous 1 (100%) eux 2 (100%)

Table 15: Pronouns in FT

As can be seen in table 15 there is typically no distinction between subject and object-forms of the pronouns. The set of pronouns proposed by the anonymous manual doesindeed appear in the other documents as well.

There are only 4 instances of reflexive forms, 2 m’, 1 nous and 1 mon corps (18).22Including 9 occurrences of li as 3sg.NOM

41

(18) Français Tirailleur (FrTir0774, Cousturier 1920:158Original Je vas tuer mon corps

StdzdGloss

Je1sg.NOM

vaFUT

tuerkill

mon1sg.POSS

corpsbody

‘I’m going to kill myself’

(Holm 1988:204) discusses instances of words for ’body’ together with a possessive formmeaning ’self’ in Atlantic creoles. There is evidence of this form in Old French (Sylvain1936:65 as cited in Holm 1988) and Chaudenson (1974:734 as cited in Holm 1988) hasdocumented its survival in some regional dialects. In a typological study of intensifiers,such as himself/herself, König and Gast (2006:224) found that the expression of these wereoften derived from words denoting body parts. It is thusly not unsurprising that there isan instance of this in FT, there is however only one.

The distinction in natural gender for 3rd person singular seems to be lost (i.e. elle’she’ - il ’he’). This has been noted for French creoles as well (Holm 1988:201). There arehowever quite few women mentioned in the material so we cannot be certain that this isreally the case. There are three instances of overt pronouns denoting women, all are foundin Lhote (1947) (19-20).

(19) Français Tirailleur (FrTir0697, Lhote 1947:304)Original Lui y en a faire cabinet partout dans le campement

StdzdGloss

Lui3sg.NOM

yenayena

fairemake

cabinettoilet

partoutall.over

dansin/at

lethe

campementcamp

‘She relieved herself all over the camp’ 23

(20) Français Tirailleur (FrTir0697, Lhote 1947:305-6)Original Alors, moi y en a dire Lieutenant et lieutenant y en a boucler lui, car luiy en a cochon beaucoup

StdzdGloss

Alors,then

moi1sg.nom

yenayena

diresay/tell

Lieutenantlieutenant

etand

lieutenantlieutenant

yenayena

bouclershackle

lui,3sg.ACC,

carbecause

lui3sg.NOM

yenayena

cochonpig

beaucoupmuch

‘So I told the lieutenant, and he shackled her, since she is a real pig/asshole’

The pronoun system of FT does indeed appear to be as the anonymous manual claim.No distinction between subject and object position and no natural gender.

42

5.2.6 Attributive possession

Attributive possession refers to constructions in which the expression of the possessum andpossessor form a noun phrase (McGregor 2009:2), such as ‘Kolbeinn’s fox’.

In French, attributive possession where the possessor is pronominal (‘my fox’) is ex-pressed primarily through possessive pronouns24 (mon renard). These pronouns agree withthe possessum in number and gender and with the possessor in person. Table 16 gives anoverview of possessive pronouns in French.

Possessum !Possessor # masculine singular feminine singular plural1sg mon ma mes2sg ton ta tes3sg son sa ses1pl notre notre nos2pl votre votre vos3pl leur leur leurs

Table 16: French possessive pronouns

When there is a nominal possessor in French this is expressed through a prepositionalconstruction (le renard de Kolbeinn ’Kolbeinn’s fox’). There is also a prepositional con-struction of à. This construction can be combined with the possessive pronouns giving aslightly different meaning; mon renard à moi - ‘my own fox’.

The anonymous manual describes three different strategies for attributive possession:(i) the possessive pronouns mon and ton for 1st and 2nd person singular respectively, (ii) aprepositional construction with pour and (iii) juxtaposition with possessum preceding thepossessor (own example: renard Kolbeinn ’Kolbeinn’s fox’).

All these constructions exist in the material, along with more prepositional construc-tions. Table 17 displays the different forms of expressing attributive possession of 1st, 2ndand 3rd singular person and nominal possessors. The most common form of the posses-sive pronouns is the etymologically masculine form (mon and ton), more on grammaticalgender can be found in section 5.2.4.

Plural possessive pronouns are excluded here, there were only two instances of these,notre - ‘our’ and juxtaposition with nous.

24In French grammar tradition they are known as adjectives, but I will label them pronouns here toavoid confusion when discussing the part-of-speech adjectives later.

43

1sg 2sg 3sg NominalTotal 86 (100%) 20 (100%) 37 (100%) 32 (100%)pronoun 79 (92%) 16 (80%) 36 (97%) 0 (0%)juxtaposition 0 (0%) 0 (0%) 0 (0%) 13 (41%)pour .... 5 (6%) 2 (10%) 1 (3%) 1 (3%)à/aux ... 2 (2%) 2 (10%) 0 (0%) 4 (13%)de .... 0 (0%) 0 (0%) 0 (0%) 13 (41%)

Table 17: Expression of attributive possession in FT

There is one example of double marking of attributive possession, both pronominal andwith the prepositional à (21). It is not clear if this should be interpreted as the typicalpossessive or the more emphasized ‘own NP’.

(21) Français Tirailleur (FrTir0491a, Hubin 1987Original moi y en a ton Noir à toi

StdzdGloss

moi1sg.NOM

yenayena

ton2sg.POSS

Noirblack

àat/to

toi2sg.ACC

‘I am your Black’

It would seem that (Anon 1916) was indeed right that the most common way for 1stand 2nd person singular is possessive pronouns, but also for 3rd person singular. Whenit comes to nominal possessors it would seem that pour... is less common than expected;in fact the construction with à/aux is more common than de, though they are both veryinfrequent.

Juxtaposition (possessum - possessor) is the most common strategy in FT when thepossessor is nominal. The construction is found in many different sources, the earliest beingfrom 1888 (Binger 1892:207). Holm (1988) notes that juxtaposition is a widespread strategyin creoles, though primarily with the order possessor - possessum. Holm also suggeststhat the juxtaposition construction with the order possessum - possessor in French creolesmight be due to the deletion of prepositional de (renard de Kolbeinn ! renard Kolbeinn).Whether or not this is the case for FT is not known, but likely.

Possessum - possessor is also the most common order among the French lexified contactvarieties in APiCS, and it is slightly more common than ’possessor - possessum’ in all thecontact varieties in APiCS (Huber and the APiCS Consortium 2013).

It would seem that attributive possession where the possessor is pronominal is mostoften expressed through possessive pronouns. Juxtaposition and the prepositional con-struction with de are equally common when the possessor is nominal. The order in thejuxtaposition is interestingly not as the anonymous manual suggests (possessor - posses-sum), it would appear that the substrate influence is perhaps less than the author(s) ofthe manual thought.

44

5.2.7 Predicative possession

Predicative possession is a clause where the ownership a certain object is predicated of apossessor (’Kolbeinn has a fox’).

Stassen (2011b) made a typological survey of the expression of predictive possession,the most frequent strategy in his sample was the habeo-verb-construction (like in Germanicand Romance languages). 26% (63/240) of the languages in that study used habeo-verbstrategy. It is however quite likely that this survey is biased and the amount of habeo-verb-languages overrepresented. The habeo-verb construction takes the possessor as subject andthe possessum as object (Kolbeinn a un renard - ‘Kolbeinn has a fox’).

The other major types in Stassen’s study are syntactically intransitive clauses whichhave the basic form of an existential clause. These constructions all features a locationalor existential predicate, often translated as ‘to be at’, ‘to be there’ or simply ‘to exist’.Collectively they make up the rest of the sample, 74% of the languages in the sample.

Stassen divides these intransitive existential clause into four subtypes: 20% of the lan-guages in Stassen’s sample has an oblique-genitive-construction, 9% oblique-locational,20% topic and 25% a conjunctional construction. The most frequent subtype in Stassensample is the conjunctional, where the possessor is the subject and the possessum accompa-nied by locational or topical marker. In the languages of his sample this marker originatesfrom a marker of simultaneity between clauses.

The oblique locational and the topical-construction is equally common in Stassen’ssample. In the oblique locational construction the possessor NP is marked for locationalrelation and the possessum is the subject. The possessum is the subject in the topicalconstruction as well, but the possessor is in this case marked as the topic. The obliquegenitive possessive construction is the least common, it is similar to the locational in thatthe possessum is the subject but there is no locational interpretation.

In Stassen’s sample we find that Bambara is coded as having a locational constructionand Hausa conjunctional. Diola-Fogny and Mòore employs habeo-verbs.

Table 18 displays the different expression of predicative possession found in the FT-material of the corpus.

Total pred poss gagner only ya only yena null41 (100%) 9 (22%) 18 (44%) 5 (12%) 5 (12%)

Table 18: Predicate possession in FT

Out of the few instances of predicative possession ya is most frequent and gagnersecond.

Ya and yena does function as the sole marker of predicative possession. Etymologicallythey both contain instances present tense 3rd person of the French habeo-verb avoir ; a.That does not, however, mean that ya and yena in FT are instances of the habeo-verb.

Ya and yena can function as locative or existential statives (see section 5.4.2 och 5.4.3).There are instances of predicative possession with no obvious verbal elements whatsoever,

45

labeled “null” in the table. Locative and existential clauses have also been found with noverbal elements.

In Stassen’s subtypes of predictive possession with intransitive existential clauses thepossessum is the grammatical subject, we do not know if this is true of the occurrencesof ya, yena or the clauses with no verbal element. The constructions are not obliquelocational, there are no locational markers on the possessor NP. As for the conjugationalsubtype, there are a few instances where ya and yena could be interpreted as relativizers.The issue of topic-marking in FT is unknown.

It is very likely that the expression of predictive possession with ya (and to a certainextent yena and null) is one of Stassen’s intransitive existential subtypes, but which one isnot clear. We need to know more about topic-marking and ya and yena as relativizers.

The anonymous manual suggests that the most common strategy for expression pred-icative possession involves the verb gagner, it would seem that ya is more common —perhaps as a result of unknown substrate influence or linguistic universal.

5.2.8 Copula relations

The term copula can be defined in many different ways. For the purpose of this study, acopular clause is a clause that expresses a stative relationship between two phrases, mostoften Nominal Phrases (NP) and Adjectival Phrases (AP), but also Prepositional Phrases(PP). This definition of copular clauses focuses primarily on the “stativeness” of the relationbetween the phrases. There are definitions of copular clauses which include the notion of‘become’ - this is not done here.

The different copular clauses found in the material have been divided into five differentcategories of copula relations.

Copula relation example formulapresentative That’s Harry! DEM = AP/NPpredicative The cat is grey NP = APequative25 Harry is a cat NP = NPlocative Harry is under the sofa NP = PPexistence Harry exists NP =

Table 19: Copula relations distinguished in this study

It is widely known that all languages need not have an overt copula for expressing theequative (Stassen 2011c), this is often called ’zero copula’. The majority of the languagesin Stassen’s study (54%, 211 out of 386) did not have the possibility for zero copula inequative clauses. Among these languages were Malgasy, Hausa, French, Mòoré, Wolof,Bambara, Maninkakan (Western) and Mandinka (Gambian).

Table 20 covers the different expressions of copula relations in the corpus.25Equative copula is similar to ‘predicate nominals’ as used by (Stassen 2011c) and copula verbs indi-

cating a Identity relation as used by Dixon (2010b:157).

46

Total zero only ya only yena être cestTotal COP 498 (100%) 169 (34%) 176 (35%) 87 (17%) 22 (4%) 34 (7%)COP Predicative 272 (100%) 93 (34%) 115 (42%) 40 (15%) 10 (4%) 8 (3%)COP Presentative 74 (100%) 26 (35%) 19 (26%) 7 (9%) 1 (1%) 20 (27%)COP Equative 100 (100%) 39 (39%) 26 (26%) 20 (20%) 6 (6%) 6 (6%)COP Locative 14 (100%) 3 (21%) 4 (29%) 3 (21%) 4 (29%) 6 (43%)COP Existence 32 (100%) 7 (22%) 7 (22%) 17 (53%) 0 (0%) 6 (19%)

Table 20: Expression of copula relations in FT

The column of être in table 20 includes être (6 occurrences), est (12), étais/t (2)(avoir/ya) été (2) and seraient (1). There are very few occurrences of être and it is likelythat many of the instances are due to authors being influenced by standard French.

Zero-marking of copula relations is very frequent in FT, as the anonymous manualsuggests. It is not found in the lexifier, nor in several of the potential substrate languages.

I have treated cest as a word in and of itself. Cest is etymologically combination ofthe demonstrative pronoun ce and the copula verb être. This decision to count cest as onenon-compositional word is based on the fact that cest can occur as a sole copula verb withother subjects and that there are no occurrences of lone ce. Out of the 22 occurrences ofcest, 10 are analyzed as subject-less and 11 have another explicit subject. Cest is found inthe work of 25 different authors, most of the occurrences of cest are before 1900.

(22) Français Tirailleur (FrTir0966 Béchet 1889:78)Original Ça c’est noirs comme nous autres

StdzdGloss

Ça3pl

cestCOP

noirblack

commelike/as

nousautres1pl

‘They are black like us’

There are several different means of expressing copula relations in FT, we need tounderstand the nature of adjectives as a part-of-speech in FT and polysemous ya and yenabetter in order to fully understand copula in FT. The analysis by Wilson (1999) and theanonymous manual that FT lacks a copula is potentially contradicted by this material.

5.2.9 ‘Being able’ — moyen and pouvoir

There are three different constructions that are given translations of ‘be able’: moyen,pouvoir and yena manière. The last one is very minor — there is only one instance. Ety-mologically moyen is a noun meaning ’means’ or ’way’, pouvoir is a French verb equivalentto English ’can’ and faire manière is a FT construction which in French means literally’do/make manner/way’ (23).

47

(23) Français Tirailleur (FrTir1080, Marceau 1911:32)Original Toubab, yena manière tout

Stdzdgloss

Toubab,white.person

yenayena

manièremanner

toutall

‘The whites can do everything’

Moyen is much more frequent, 70 out of 79 instances of ‘be able’ are expressed. Theremaining 8 cases are expressed by the verb pouvoir (7 pouvoir, 1 peut). Pouvoir occursmore in the older material whereas moyen mainly appears after 1900.

5.2.10 Summary: function seeking form

The main strategy of expression polar interrogation appears to be through intonation,standard negation is expressed by pre-verbal pas and non-standard negation is very similar(or even identical) to the lexifier.

There is little reason to assume that FT has a productive grammatical gender systemand nor is there a difference in natural gender in the pronoun system. There is no distinctionbetween subject and object position of the pronouns. In order to express attributivepossession FT employs possessive pronouns if the possessor is pronominal and singular.If the possessor is nominal attributive possession can be formed either with juxtaposition(possessum - possessor) or prepositional construction with de (but there is evidence ofother preposition as well).

The most frequent verbal element in clauses which predicts possession is ya and secondlygagner. In the study by Stassen (2011b), it is common for predicative possession to beexpressed by an intransitive clause involving a verb which can denote existence, this couldbe true of FT. The important difference lies in the marking of the possessum, which inStassen’s analysis is the subject of the clause. We do not know if this is true of FT as well.

There is indeed zero-marking of copula-clauses, like Wilson and the anonymous manualsuggests, but copula relations can also be marked by ya and yena. In order to fullyunderstand copula relations we need to define whether or not adjectives constitutes adistinct part-of-speech.

The notion of ‘be able’ can be expressed by moyen.

5.3 Form seeking function

This section is devoted to the exploration of the function of certain very frequent forms inthe material, the mirror image of the previous section.

5.3.1 Drop of overt subject

There are clauses that contain an explicit formal subject, but this subject carries no se-mantic meaning. These are called “dummy subjects” or “expletives”. The most commonly

48

example of this are statements about the weather, such as it rains and existence there

are no ghosts. Some languages, like English require an explicit subject pronoun here evenwhen it contributes no meaning, other languages do not require these semantically emptyformal subjects.

In French it is obligatory to mark person agreement with the subject on the verb.French is coded as having ’obligatory pronouns in subject position (Dryer 2011a). It waspossible, however, to omit pronominal subjects in Old French26 (Adams 1987:2) and inLatin. There is still the possibility to omit subjects in sentence-initial position in French incertain registers and early production (Haegeman 1997:236). Studies on English (Nariyama2004) suggest that while English is not pro-drop, it is indeed possible to omit pronominalsubjects in casual conversation.

There are many instances of subject drop in the FT-material, and the phenomenonis not unique to any certain time period or author of document. Table 21 displays theoverall occurrences of drop of overt subject. ‘Drop of dummy’ stands for instances whereit is clear that the omitted subject falls within the category of dummy subjects. ’NullSubject’ stands for instances where there is a semantic subject. If it is unclear in the FTclause whether there is null subject (3sg) or drop of dummy that clause has been annotatedfor null subject. There is a division between dynamic and stative clauses, and a furtherdivision of the stative clauses into different copula-subtypes.

Total All drop Null subject Drop of dummyTotal 1 737 (100%) 318 (18%) 263 (14%) 55 (4%)Indep 1 521 (100%) 289 (19%) 216 (14%) 73 (5%)Sub 208 (100%) 25 (12%) 18 (9%) 7 (3%)Dynamic 822 (100%) 115 (14%) 104 (12%) 11 (1%)All COP 498 (100%) 127 (26%) 95 (19%) 32 (6%)COP existence 29 (100%) 22 (76%) 1 (3%) 21 (76%)COP locative 17 (100%) 2 (12%) 0 (0%) 2 (2%)Only ya 192 (100%) 54 (28%) 43 (22%) 11 (6%)Only yena 97 (100%) 24 (25%) 7 (7%) 17 (18%)

Table 21: Drop of overt subject

Drop of subject is quite common in the FT-material, primarily in copula clauses. Im-perative clauses often lack overt subject and have not been included in table 21, 63% ofthe imperative clauses had no explicit subject.

In 83 of the clauses with subject drop the subject was identical to that of a immediatelypreceding clause, making the task of the listener to identify the referent much easier. It ispossible that there are more instances where there is contextual information available tothe listener, we cannot know for sure. Here is one example of this type of subject drop.

26Old French is often defined as spoken 1100-1300.

49

(24) Français Tirailleur (FrTir0565, Diallo 1926:86)Original Moi pas bien connaître dire français, [...] mais connais lire français danscœur à toi

StdzdGloss

Moi1sg.NOM

pasNEG

biengood

connaîtreknow

direspeak

françaisFrench

[...][. . . ]

maisbut

connaîtknow

lireread

françaisFrench

dansin

cœurheart

àat

toi2sg.ACC/DAT

‘I don’t speak French well, [...] but I can read the French of your heart ’

There are 10 instances of the copula cest where there is no overt subject, these havebeen coded as instances of subject drop (see section 5.2.8). If we were to analyze cest ascontaining a pronoun there would be 328 cases of drop of over subject (incl ’dummies’),i.e.19% of all clauses instead of 18%.

Which subjects are most often dropped then? If we consider the translation and sur-rounding clauses, it seems like most deleted subjects are of 3rd person singular. Table 22displays the amount of explicitly marked subjects and the null subjects, i.e. drop of overtsubject that is not a dummy.

1sg 2sg 3sg 1pl 2pl 3plExplicit subject 309 204 336 71 9 40Null subject 54 26 140 12 0 28

Table 22: Null subject distributed over person and number

We cannot be certain of the conditions in which the subject can be dropped. It couldbe an influence of older varieties of French or a result of the spoken nature of FT. Inmost contact varieties with a French lexifier in the APiCS-sample it is obligatory to havean overt pronominal subject if the referent is highly activated. (Haspelmath and theAPiCS Consortium 2013)27. The issue of drop of overt subject is not discussed in previousliterature on FT, it needs to be investigated further.

27Once more, these varieties are most often defined as creoles.

50

5.3.2 Polysemous form ya

Figure 8: Advertisement for chocolate powder. c�1915 Giacomo de Andreis.

One of the most striking elements of FT is ya. In 1915 the company Banania used the slogany’a bon to market their chocolate power in France (figure 10). Banania’s advertisement forchocolate powder still features a West African soldier, a Tirailleur Sénégalais. However,today the character speaks standard French.

Ya occurs in 39% of the strings in the corpus, from the material from 1860’s all theway until the 1950’s. In 198 clauses (11%) there is no other obvious verbal element thanya (see table 23).

The origin of ya is the il y a-construction in French, which means ‘there is’. Thisconstruction has two functions: firstly existence/location (25) and secondly ’passed time’(26).

(25) French, il y a, existence (own example)

il3sg.NOM

ythere

ahave.PRS.3sg

una

renardfox

dansin

ton2sg.POSS

jardingarden

‘There’s a fox in your garden’(26) French, il y a, passed time (own example)

KolbeinnKolbeinn

estbe.PRS.3sg

néborn

il3sg.NOM

ythere

ahave.PRS.3sg

trente-et-unthirty.one

ansyears

‘Kolbeinn was born thirty-one years ago’

51

The French construction must be formed with the 3rd person pronoun il in both in-stances. As we saw in section 5.3.1, it is not common (or even possible) to drop overtsubjects in French. The pronoun il is not very common in FT overall , the most commonform for the 3rd person pronoun is lui, out of the 204 occurrences of 3sg subject positionthere were only 17 of il (see section 5.3.5).

Most of the occurrences of ya in the corpus material are not formed with il, many evenlack overt subject all together. The most common position of ya is directly before thepredicate (27), be it verbal or adjectival. This is true of 98% of the occurrences of ya.

(27) Français Tirailleur (FrTir0313, Cousturier 1920:55)Original quand moi y a parler ça. . .

StdzdGloss

Quandwhen

moi1sg

yaya

parlerspeak

çaDEM

‘when I said that... ’

If there are negators (pas, jamais, rien, plus) they occur after ya and before the mainverb. This is identical to the position of negators in French with respect to the finiteauxiliary verbs (see section 5.3.2.).

All instances of ya, both when occurring before other verbs and when it appears alone,were tested for correlations with different potential relevant tense-aspect-contexts. TheTA-contexts have been inferred from the context and translations. There was no clearcorrelation, it would seem that there are no overt markers of TA in FT.

Table 23 shows how ya co-occur with different functions and contexts.

52

Total ya only yaTotal 1 737 (100%) 682 (39%) 198 (11%)Dyn 823 (100%) 297 (36%) 0 (0%)All stative (incl Pred. Poss & COP) 908 (100%) 383 (42%) 197 (22%)All COP 499 (100%) 183 (37%) 176 (35%)COP Equative 100 (100%) 26 (26%) 26 (26%)COP Predicative 272 (100%) 121 (44%) 116 (43%)COP Presentative 74 (100%) 21 (28%) 19 (26%)COP Locative 14(100%) 4 (29%) 4 (29%)COP Existence 32 (100%) 7 (22%) 7 (22%)Pred. Poss. 40 (100%) 22 (55%) 18 (45%)Subj: 3sg 476 (100%) 209 (44%) 89 (19%)Subj: N 318 (100%) 171 (54%) 33 (10%)Null subject 263 (100%) 130 (49%) 45 (17%)Drop of dummy 55 (100%) 24 (44%) 12 (22%)PST 279 (100%) 99 (35%) 8 (3%)PROG 79 (100%) 26 (33%) 4 (5%)FUT 113 (100%) 33 (29%) 5 (4%)

Table 23: Ya in FT

The occurrence of ya in the corpus suggests that it is highly polysemous, but it doesnot seem to have a temporal or aspectual function. Ya does not occur alone as a dynamicverb, but precedes dynamic verbs — perhaps as a marker of finiteness as Wilson suggests.When alone ya has a stative meaning of copula or possession.

There is only one instance that could be interpreted as the ’passed time’-constructionof French:

(28) Français Tirailleur (FrTir1180, Anon 1899:243Original [...] moi pas content il y a vingt-deux jours

StdzdGloss

moi1sg.NOM

pasNEG

contentbe.happy

yaya

vingt-deuxtwenty-two

jourdays

‘I wasn’t happy twenty-two days ago/It’s been twenty-two days since I was happy’

Ya also forms a near-obligatory/obligatory part of two verbal constructions. ya besoin’need’ and ya moyen ‘be able/work’. Table 24 shows the amount of strings that containya and yena in stative and dynamic clauses; and clauses with besoin ‘need’ and moyen ‘beable to’ present.

53

Total ya only yaTotal 1 737 (100%) 682 (39%) 198 (11%)All stative (incl Pred. Poss & COP) 908 (100%) 383 (42%) 197 (22%)Dynamic 823 (100%) 297 (36%) 0 (0%)besoin 55 (100%) 48 (87%) 0 (0%)moyen 78 (100%) 61 (78%) 0 (0%)

Table 24: Ya besoin and ya moyen.

After examining the instances of ya in the anonymous manual, Wilson (1999) drawsthe conclusion that it is a marker of finiteness and that it is also a stative verb denotingexistence. It would seem that ya does indeed occur mostly before the predicate, be thisverbal or non-verbal, however, as can be seen in the table above, ya does occur alonewithout any other verbal elements.

In the case of predicative and presentative copulas, it is possible to argue that ya isnot the only verbal element. If we make no distinction between adjectives and verbs, butrather view adjectives and verbs as one part-of-speech in FT, then the instances of yabefore these adjective-verbs is not very different from ya before other verbs. This could bedescribed as a predicate marker or of finiteness.

This is, however, not an appropriate analysis of the possessive and existential/locationalinstances though, we have no evidence that there is any other element in those clausecapable of carrying the possessive or existential/locational meaning.

Previous literature has suggested that ya is not a copula (and that there is no overt cop-ula in FT) but a marker of finiteness and sometimes also stative verb of location/existence(Wilson 1999). In our corpus material we find instances of ya as a marker of predicativepossession (perhaps as a verb of location/existence, see section 5.3.7) and equative copula.

It is possible that the function of denoting location/existence is closely related to thatof predicative possession. Many languages employ a intransitive existential construction,including presumably much influential Bambara.

Our corpus suggests that ya is highly polysemous. The following functions are attestedin several different sources:

• equative copula (26 occurrences)

• stative verb predicating location or existence (11 occurrences)

• stative verb predicating possession (18 occurrences)

• before adjectives, potentially predicate marker or copula verb (116 occurrences)

• before dynamic verb, potentially as predicate marker (297 occurrences)

• obligatory part of verbal construction (109 occurrences)

54

5.3.3 Polysemous form yena

Another highly frequent element of FT is yena. Yena stems from the French construction ily en a which also means ’there is’. Il y en a is very similar to il y a, the important differencelies in en. en is a pronoun which denotes 3rd person accusative or the prepositionalconstruction of de + NP. The habeo-verb avoir (il y en a) requires an object, this meansthat il y en a can be an independent clause in standard French in contrast to il y a whichis not an independent clause.

Il y en a is a relatively uncommon construction; in a corpus of modern spoken French(Debrock et al. 2001) it occurred 564 times in a collection of 902 756 words (cf. y a: 6201occurrences).

Table 25 shows how yena co-occur with different functions and contexts.

Total yena only yenaTotal 1 737 (100%) 249 (14%) 97 (6%)Dyn 823 (100%) 103 (13%) 0 (0%)All stative (incl Pred. Poss and all COP) 908 (100%) 145 (16%) 96 (11%)All COP 499 (100%) 93 (19%) 87 (17%)COP Equative 100 (100%) 22 (22%) 20 (20%)COP Predicative 272 (100%) 43 (16%) 40 (15%)COP Presentative 74 (100%) 8 (11%) 7 (9%)COP Locative 14 (100%) 3 (21%) 3 (21%)COP Existence 32 (100%) 17 (53%) 17 (53%)Pred. Poss. 40 (100%) 8 (20%) 5 (13%)Subj: 3sg 476 (100%) 53 (11%) 30 (6%)Subj: N 318 (100%) 105 (33%) 52 (16%)PST 279 (100%) 48 (17%) 7 (3%)PROG 79 (100%) 16 (20%) 3 (4%)FUT 113 (100%) 19 (17%) 1 (1%)

Table 25: Yena in FT

As table 25 shows yena occurs often in combination with other verbal elements, but itcan mark different copula relations and also predictive possession on its own. The corpusmaterial does not seem to suggest any temporal or aspectual function of yena, but this isneeds to be further investigated.

French makes use of a habeo-verb to express predicative possession, avoir. This verbforms part of the il y en a that is the origin of yena. We cannot exclude the possibly thatthis is the reason that yena can mark predicative possession. It is however also possiblethat yena in its function as a marker of location/existence can mark predicative possessionas well. Many languages in the study by Stassen (2011b) made use of an intransitiveconstruction involving an existential verb for expressing predicative possession (see section5.3.7).

55

The anonymous manual states that yena is used in a periphrastic construction to ex-press adjectival meaning (29) and suggests that this is a result of substrate influence. Theseexamples are however only found in the anonymous manual and not in any of the othersources.

(29) Français Tirailleur (FrTir1267, Anon 1916)OriginalTirailleur y en a pas bon, lui pas obéir

Stdzdgloss

Tirailleurtirailleur

yenayena

pasNEG

bon,good

lui3sg.NOM

pasNEG

obéirobey

‘A bad soldier doesn’t obey’

This appears to be the case for yena as a relativizer as well, that function is almostentirely restrict to the utterances from the anonymous manual.

Our corpus suggests that yena, like ya is highly polysemous:

• as equative copula (20 occurrences)

• stative verb predicating location or existence (20 occurrences)

• before adjectives, potentially predicate marker or copula verb (40 occurrences)

Yena is similar to ya but less frequent. The use of yena in predicate possession might,like ya be an instances of Stassen’s intransitive existential possession constructions, butthere are too few examples.

The issue of yena as a relativizer needs to be further investigated.

5.3.4 Gagner — to get, become or have?

In Standard French gagner means ‘win’, ‘gain’, ‘acquire’ or ‘get’. There is in total 65occurrences of gagner in the corpus. In the majority of these occurrences gagner appearsto be very similar to the etymological lexical form in French. 11 cases have been translatedas ‘become’ and 9 as ‘have’.

Most of the cases of gagner as ‘have’ are found in the anonymous manual, suggestingthat it is nor necessarily true for all FT. The ‘become’-meaning however is found in severalsources, all of them after 1916 (31).

(30) Français Tirailleur (FrTir0138, Marie-Victoria 1921:499)Original Toi, ma sœur, ya gagné Commandant

Stdzdgloss

Toi,2sg.NOM

masœur,my.sister

yaya

gagnerbecome

Commandantboss

‘You sister, you shall be the boss’

56

The function of gagner appears to be subject to change and not restricted to possession.

5.3.5 Wanting, liking and being happy — content and vouloir

There are two strategies for expressing ’want’ in FT; content (18 occurrences) and vouloir(29). There are quite few occurrences, but it seems as if vouloir is older occurring primarilybefore 1900 and content in the sense of ’want’ is more recent.

Etymologically, content is an adjective meaning ’satisfied/happy’ and content occurs20 times in the sense of ’happy and 20 as the verb ’like’. One might expect the sense of’enjoy/like’ to be more recent, but (31) dates to 1899.

(31) Français Tirailleur (FrTir0660, Montrozier 1902:155)OriginalMahomet pas content

Stdzdgloss

MahometMohamet

pasNEG

contentlike/enjoy/be.happy

‘Mohamet wouldn’t like that’

Like gagner, content seems to be subject to change and perhaps pushing out the olderform of vouloir in the function of ‘want’. Further investigation into function of content asa verb meaning ‘be happy’ might shed light on the statue of adjectives as a part-of-speechin FT.

5.3.6 Reduplication

Reduplication is defined by Rubino (2011) as the “repetition of phonological material withina word for semantic or grammatical purposes”. Languages can employ full reduplication(the repetition or words or roots) but it is also possible to have partial reduplication (therepetition of smaller segments such as phonemes).

In order for a language to be classified as having reduplication this process needs to beproductive, i.e. applicable to a set of open class words and still in current use. In Rubino’ssample 277 languages have partial reduplication, 35 full reduplication and 56 no productivereduplication.

There are many possible functions of reduplication, Rubino lists a plural number, con-tinued aspect and case among many others.

Corne (1999:201) suggests that there is reduplication in FT and that it marks intensityor continuity. Rubino lists intensity and continuity as possible functions of reduplication.There is no mention of reduplication in the thesis on FT by Wilson.

It has been claimed that reduplication is a common phenomenon in pidgins, as it isin many creoles. Bakker (2003) challenged that perception and successfully proved thatreduplication is not as common in pidgins as it had previously been described.

57

While there is not productive reduplication in French, there is productive reduplicationin some of the substrate languages - most notably there is full and partial reduplication inHausa, Maninkakan (Western) and Mòoré. There is also reduplication in Wolof, but onlyof full words or roots.

Let us examine what evidence of reduplication there is in the corpus. If we considerall instances where there is any form of repetition of a preceding element (be it an entireword or a single phoneme) we end up with 35 potential candidates of reduplication. 12 ofthese are found in the anonymous manual from 1916.

The first examples that we can exclude is the onomatopoetic tam-tam found in Fr-Tir0395a (Binger 1892:120-1) and FrTir0395a (Claudel-Hubin 1991).

Secondly, there are instances which are loans from the West African languages and thatwe should exclude from evidence of productive reduplication (table 26).

gris-gris amulet (Anon 1915:152)grigri medicine (Béchet 1889:86)bagabaga termites (Béchet 1889)ménéméné black ants (Béchet 1889:78)

Table 26: Loans with reduplication

Thirdly, we have the issue that reduplication is defined as occurring within a word.However if we trust the segmentation of material into words as done by the authors theremaining instances of reduplication are not word-internal. The concept of a word is centralin distinguishing ’reduplication’ from ’word iteration/repetition’.

Gil (2005:33) discusses this problem and notes that the repetition of words can havesimilar functions to reduplication, such as intensity of iterative aspect. Gil has set up a setof criterions for distinguishing reduplication from word repetition/iteration. Unfortunatelyseveral of these criteria cannot be applied to the FT-material since they rely on phonologicaldistinctions and we do not have access to spoken data. Many of the instances of potentialreduplication fulfill criteria of both Gil’s repetition and reduplication

According to Gil it is very unusual, but not unheard of, that reduplication producesmore than 2 copies. In the FT material we have three instances of three copies. Theseare most likely instances of word repetition. Another criteria that Gil proposes is that theinput in the reduplication process should be one word (or less), which means that we canrule out Femme tout nu, tout nu! (Nordenck 1886:276) since it is one of the clear examplesof multiple word-input.

12 of the remaining 28 instances are found in the anonymous manual. All of themexpress with spaces in between the words and what can be best described as intensification(32) or continuity.

58

(32) Français Tirailleur (FrTir0107, Cousturier 1920:152Original trois verres petits, petits, petits

StdzdGloss

troisthree

verreglass

petitsmall

petitsmall

petitsmall

‘Three very small glasses’

Now, it is not possible to say with great certainty that these instances are reduplicationor word repetition/iteration that marks intensity/emphasis/continuity. However, theseforms are very infrequent, mainly restricted to the anonymous manual and since it ispossible for word repetition to have similar function as reduplication the meaning is notcrucial to the distinction.

It would appear that FT does not have productive reduplication, it is at least not asvery frequent feature.

5.3.7 Summary: form seeking function

Subjects are sometimes omitted in FT, 18% of the material. In many of these instances(83 to be precise) the subject was the same as in the preceding clause, suggesting thatit is possible to omit subject under identity (what is known as SS in switch-referencelanguages). We cannot know the nature of subject drop in FT for certain, we have toolittle information. The literature suggests that it could be influence from French or afeature of casual spoken interaction.

Ya has many functions, one of these may be as a predicate marker preceding dynamicverbs and a stative verb denoting possession, location/existence and equative copula. Inorder to fully understand ya we need to understand adjectives as a part-of-speech — if it isnot distinct from verbs then ya could be analyzed as a predicate marker in more instances.

Yena has previously been analyzed as a stative verb and relativized. Yena as relativizedis not attested in enough sources. The function of yena is, like ya, dependent on ourdefinition of adjectives.

Gagner ‘win/acquire/get’ is often used in FT in the sense of ‘get’ or ‘have’, but in morerecent material also as ‘become’.

Content in FT can mean ‘want’, ‘like’ or ‘be happy’. The lexifier form of vouloir forexpressing ‘want’ is more common in older material, it would seem that both gagner andcontent are subject to change.

One can argue that reduplication is present in FT, but it is not a frequent feature andis much more frequent in the anonymous manual than any other source.

59

5.4 Notes on diachrony

5.4.1 Order of modifier and head

Another potential trend is the order of modifier and head within a Noun Phrase. Thereare slightly more occurrences of the head preceding the modifier in more recent material.

However, the construction with a postposed numeral that we find in the anonymousmanual seems to be restricted to the manual. In Bambara and many other West Africanlanguages, the numeral is postposed to the noun (Dryer 2011f). It might be that the authorof the anonymous manual has exaggerated the influence of Bambara in FT.

5.4.2 Aspectual fini?

There is one instance of fini in an unusual position (33) that might indicate a potentialstrategy for marking completive, perfective or perfect aspect.

(33) Français Tirailleur (FrTir0506b, Maran 1921:96)Original Moi y’en a croire lui crévé fini,

Stdzdgloss

Moi1sg.NOM

yenayena

croirebelieve

lui3sg.ACC

créverdie

fini,finish

‘I think he has passed out’

There are other examples of fini in the sense of ’finish doing/being’, but this exampleis different. Judging from the post-verbal position of fini, the context and translation itwould seem that it is possible to interpret it as aspectual.

Bybee et al. (1994:67) have noted that one possible origin for markers of Resultativity,Perfective or Perfect aspect is indeed verbs of ‘finish’. In a typological survey of the Perfectby Dahl and Velupillai (2011) 21 languages (20% of all that had Perfect) had a markerthat was derived from a word meaning ‘finish or ‘already’. There is also evidence of fin(i)as marker of completive aspect in several French creoles (Holm 1988:162-163). There isalso a post-verbal construction of the verb ’finish’ together with the infinite marker thatcan mark completive aspect in Bambara (Bird et al. 1977 as cited in Holm 1988:163). Ifthere is a completive/perfective/perfect construction in FT which originates from a verbmeaning ‘finish’ it could be a result of a linguistic universal or substrate influence.

Example 33 is however the only example of this potentially aspectual fini in FT.

5.4.3 Pre-verbal qui/que

There is one form that shows up in the early material (34), pre-verbal qui/que (labeledkV in the standardized material). Other occurrences of qui/que appear, as in the lexiifer,clause-initially28, but this is not the case here.

28This means that the adjectival periphrastic constructions as suggested by the anonymous manual aremost likely excluded.

60

(34) Français Tirailleur ( FrTir0553, Nordenck 1886:286Original Moi qu’a conni un capitaine

StdzdGloss

Moi1sg.NOM

kVkV

connuknow

unun

capitainecapitaine

‘I know/knew a captain’

There 15 occurrences of this, 13 of these are from the 1880’s, suggesting that this wasperhaps an early construction that didn’t last. It has been noted by seven different authors.The production of the most recent one is dated to 1931 (35).

(35) Français Tirailleur ( FrTir0352, Laverrière 1932:147Original Père, vite, la chapelle qui brûle

StdzdGloss

Père,father

vite,fast

laDEF

chapellechapel

kVkV

brûleburning

‘Father, quick, the chapel is on fire!’

This form always occur before the predicate and it is not found more often in clausesthat have a past tense-reading. The form might be similar in function to ya and yena.There is a possibility that the form marks focus, but this is purely speculative.

5.4.4 Summary: notes on diachrony

It can be hard to distinguish between actual historical trends in FT and authors who aremore less influenced by French.

There are few potential authentic trends visible in the corpus, the use of moyen for ‘beable’ as opposed to pouvoir (section 5.2.9), content as ‘want’ instead of vouloir (section5.3.5) and the order of modifier and noun (section 5.4.1).

The issue of aspectual fini (section 5.4.2) can only be speculation at this point, butdoes have interesting theoretical implications. As for the pre-verbal qui/que it is possiblethat it is a marker of focus or predicate, but it would seem that the form disappeared earlyon.

The order or elements, not only modifier and head, but perhaps also of the elements ofthe finite clause would be an interesting issue to pursue further.

61

6 Discussion

There are many more features of FT that should be further investigated (see section 3.2.3),and we are, as always, in need of more data nor only on FT but also on pidgins in general.

One reoccurring issue is the spoken nature of FT, it was rarely written. We need toknow more about the spoken French of that time in order to fully understand FT. Theconditions that play into spoken conversation must be taken into account when workingwith pidgins. Omission of subject is one such example where the spoken modality couldbe relevant, when speaking it is possible to make use of many other means of transmittingmeaning (shared knowledge, pointing, etc). These additional means of communication isnot possible in the written form to the same extent.

We expect pidgins to have a restricted lexicon and many polysemous words, whichcould be reflected in a low TTR. We also expect spoken material to have a lower TTRthan written. The standardized TTR of this corpus is 27%, is this necessarily unusuallylow? More comparisons between pidgins and spoken data of languages lacking inflectionalmorphology is needed.

There are many things we do not yet know about spoken conversation. As our knowl-edge of spoken language grows, so will our understanding of the mechanisms of pidginiza-tion.

It has been stated that FT lacks overt copula (Anon 1916 and Corne 1999). It is notclear what exactly they mean by ’copula’, I assume that they primarily refer to equative,presentative and predicative copula relations (excluding locative and existential). In orderto answer this question we must first understand ya and yena and if there is indeed adistinct part-of-speech of adjectives.

In the case of ya it is possible to analyze it as a marker of finiteness or the predicateinstead of copula verbs29. This, in combination with the high number of zero forms ofcopula in predicative copulas, brings up the issue of whether or not it is relevant to makea distinction between adjectives and verbs in FT.

An argument for the lumping together of these two categories would be the lack of verbalmorphology and syntax, there are few formal differences between (what is etymologically)adjectives when they occur in predicative position and what is defined as verbs in French.They can both be preceded by ya or yena.

We could define adjectives in FT as ’verby’ when they occur in predicate position, usingthe terminology of Wetzer (1996). In the typological study of the expression of predicativeadjectives Stassen (2011a) found that verbal encoding was more common than non-verbal.In Wolof what can be defined as predicative adjectives are expressed verbally, Mandinka,Mòoré and Bambara has both strategies Stassen (2011a). Verbal encoding of what is oftenadjectives in IE is also common in creoles (Holm 1988:176-177).

An argument for the division of verbs and adjectives into two different parts-of-speechwould be that words that are etymologically adjectives can occur within an NP, verbscannot. A second argument for the division is the occurrence of être and cest, if adjectives

29This is not an exhaustive definition of ya and yena, but it is one potential explanation of their function.

62

and verbs belong to the same category why can they be preceded by another verbal element?Another argument against lumping verbs and adjectives into the same part-of-speech is

that adjectives can modify the head of an NPs in FT, and in that they do differ from verbs.The status of adjectives as a part-of-speech in pidgins need to be further investigated ifwe are to understand the grammar of FT in general and the nature of ya and yena inparticular.

Many of the predictions by the anonymous manual and Delafosse are supported by thecorpus material. However, there are features that appear exclusively in the manual, suchas post-nominal numerals and the periphrastic adjectival constructions with yena. Thesefeatures are both similar to structures that exist in Bambara and Wolof, two languages thatthe author(s) of the manual most likely knew were spoken by the soldiers. The author(s)of the manual might have exaggerated Bambara’s influence (Van Den Avenne (2012:258)even calls the FT of the manual a calque of Bambara). There is also the possibility thatthese features occurred more often in the variety of FT that the manual portrays.

The results of this study suggests that the influence from West African substrate lan-guages on these features is minor, it would seem that the overall reduction of the pidginiza-tion is the more important. It has been suggested that structures that are typologicallycommon are more likely to show up in reduced language varieties such as pidgins. Ex-amples of these structures are: the expression of polar interrogation through intonationand pre-verbal negation. The reason why they show up in pidgins might be due to thefact that (i) they are common in the world’s languages (i.e. more likely to occur in theinvolved languages), or (ii) they are “cognitively less complex” and therefore preferablein these contact situations. We need more empirical investigations of pidgin corpuses tofurther understand this.

It is worth noting that much of the new material of FT that hasn’t been brought uppreviously in the literature does behave like the manual. The structures of FT that havebeen described in this thesis are found in the majority of the sources, this suggests thatFT was conventionalized enough to be labeled a ‘pidgin’.

63

7 Conclusions

In this corpus study we can discern the following of the nature of FT:

• polar interrogation is, most likely, expressed through intonation (see section 5.2.1)

• standard negation is expressed through pre-verbal particle pas (see section 5.2.2)

• non-standard negation is expressed identical or very similar to the lexifier French (seesection 5.2.3)

• there is no productive system of grammatical gender (see section 5.2.4)

• there is, most likely, no distinction of natural gender in 3rd person pronouns (seesection 5.2.4 and 5.2.5)

• the pronoun system consists primarily of: moi, toi, lui, nous, vous and eux (seesection 5.2.5)

• when the possessor is pronominal attributive possession is expressed by possessivepronouns (see section 5.2.6)

• when the possessor is nominal attributive possession is expressed through juxtaposi-tion and prepositional constructions (see section 5.2.6)

• when attributive possession is expressed through juxtaposition the relative order ispossessum - possessor

• there is a a highly polysemous morpheme ya that can occur as marker of predicativepossession, existential/locational verb, equative copula and possible as a predicatemarker and/or predicative copula (see section 5.3.2)

• there is a second a highly polysemous morpheme yena that can occur as a existen-tial/locational verb and potentially also as a copula (see section 5.3.3)

There are instances of potential change over time, but nothing can be said for certainas for the diachronic development of FT (see section 5.4).

It is very likely that comparisons with research on spoken language can contributetremendously to our understanding of the lexicon and structure of pidgins and pidginiza-tion, in particular corpuses of spoken utterances of languages lacking inflectional morphol-ogy. We need more empirical studies of pidgins to understand their nature.

The features of FT that were found in this thesis are documented in the works of manydifferent authors. This suggests that FT did indeed exist and that it was stable enough forus to label it a ‘pidgin’. As is the case with all studies of language, there is great internalvariation and the different language varieties are merely overlapping in enough many areasfor linguists to be satisfied with calling it one entity.

64

References

Adams, M. (1987). From Old French to the theory of pro-drop. Natural Language &Linguistic Theory, 5.

Alsenoy, L. V. (2011). Negative expressions involving indefinites in African languages.Papers of the Linguistic Society of Belgium, 6.

Anon (1899). Par ici, par là. Journal de la maison, 9.

Anon (1915). Aux Dardanelles (février-mars 1915). La Revue de Paris, 22:pp 135–159.

Anon (1916). Le français tel que le parlent nos tirailleurs Sénégalais. Imprimerie MilitaireUniverselle L. Fournier, Paris.

Augouard, P. (1905). 28 années au Congo, volume 1. Société française d’imprimerie et delibrairie, Poitiers.

Baker, P. (2006). Using Corpora in Discourse Analysis. Contiuum, London.

Bakker, P. (1997). A language of our own: the genesis of Michif, the Mixed Cree-Frenchlanguage of the Canadian Métis, volume 10 of Oxford Studies in Anthropological Lin-guistics. Oxford University Press. PhD U Amsterdam 1992.

Bakker, P. (2003). The absence of reduplication in pidgins. In Kouwenberg, S., editor,Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages,Westminster Creolistics Series 8, pages 37–64. Battlebridge.

Baratier, A. (1912). Épopées Africaines. Arthème Fayarde, Paris.

Barret, P. (1888). L’Afrique occidentale: la nature et l’homme noir, volume 2. Challamel,Paris.

Béchet, E. (1889). Cinq ans de séjour au Soudan français. Librairie Plon, Paris.

Bellaigue, B. (2009). Indochine. Publibook, Paris.

Biasini, E. (1995). Grands travaux: de l’Afrique au Louvre. Éditions Odile Jacob.

Bickel, B. and Nichols, J. (2011). Fusion of selected inflectional formatives. In Dryer,M. S. and Haspelmath, M., editors, The World Atlas of Language Structures Online.Max Planck Digital Library, Munich.

Binger, L. G. (1892). Du Niger au Golfe de Guinée par le pays de Kong et le Mossi.Hachette, Paris.

Biondi, J.-P. (1987). Saint-Louis du Senegal: memoires d’un metissage. Denoel, Paris.

65

Bobe, B. (2007). Utilisation et écriture des titres de civilité. http://www.culture-generale.fr/.

Bouquet, L. and Hosten, E. (1917). L’Epopée Sénégalaise sur l’Yser. L’Illustration 3375.

Broch, I. and Jahr, E. H. (1981). Russenorsk - et pidginspråk i Norge. Novus, Oslo.

Bybee, J. and Dahl, Ö. (1989). The creation of tense and aspect systems in the languagesof the world. Studies in Language, 51-103.

Bybee, J., Perkins, R., and Pagliuca, W. (1994). The evolution of grammar: Tense, aspect,and modality in the languages of the world. The University of Chicago Press, Chichago.

Chafer, T. (2002). The end of empire in French West Africa: France’s successful decolo-nization? Berg, Oxford.

Claudel-Hubin, M. (1991). La ferme du Blanc. En Pays Mossi.

Conombo, J. (1989). Souvenirs de guerre d’un ”tirailleur sénégalais”. l’Harmattan, Paris.

Corbett, G. G. (2011). Number of genders. In Dryer, M. S. and Haspelmath, M., editors,The World Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Corne, C. (1999). From French to Creole - The development of new vernaculars in theFrench colonial world. Westminister Creolistics Series. University of Westminister Press,London.

Cousturier, L. (1920). Des inconnus chez moi. Éditions de la Sirène, Paris.

Dahl, Ö. and Velupillai, V. (2011). The perfect. In Dryer, M. S. and Haspelmath, M.,editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

David, D. (1971). Pidgin and creole languages. In Hymes, D., editor, Pidginization andCreolization of Languages. Cambridge University Press, Cambridge.

de Mandat-Grancey, E. (1900). Au Congo (1898): impressions d’un touriste. Plon, Paris.

Debrock, M., Mertens, P., Truyen, F., and Brosens, V. (2001). Etude Linguistique de laCommunication Parlée (ELICOP).

Delafosse, M. (1904). Parlés à la côte D’Ivore, chapter VIII Les langues étrangères, pages254–286. Leroux, Paris.

Delafosse, M. (1931). The Negroes of Africa. Societé d’éditions géographiques, maritimeset coloniales, Paris.

Desjardins, R. (1925). Avec les Sénégalais par de là l’Euphrate. Calmann-Lévy, Paris.

66

Dewitte, P. (1990). La rouge et la negre. In Presse et memoire: France des etrangers,France des libertes. Mémoires génériques : Editions ouvrières, Paris.

Diallo, B. (1926). Force-bonté. Rieder et Cie, Paris.

Diarra, B. (1927). Réponse d’un ancien tirailleur sénégalais à M. Paul Boncour sur sa loimilitaire. La Race Nègre, 1.

Dixon, R. M. W. (2010a). Basic linguistic theory. Vol. 1, Methodology. Oxford UniversityPress, Oxford.

Dixon, R. M. W. (2010b). Basic linguistic theory. Vol. 2, Grammatical Topics. OxfordUniversity Press, Oxford.

Dryer, M. S. (2011a). Expression of pronominal subjects. In Dryer, M. S. and Haspelmath,M., editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

Dryer, M. S. (2011b). Genealogical language list. In Dryer, M. S. and Haspelmath, M.,editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

Dryer, M. S. (2011c). Negative morphemes. In Dryer, M. S. and Haspelmath, M., editors,The World Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Dryer, M. S. (2011d). Optional double negation in svo languages. In Dryer, M. S. andHaspelmath, M., editors, The World Atlas of Language Structures Online. Max PlanckDigital Library, Munich.

Dryer, M. S. (2011e). Order of negative morpheme and verb. In Dryer, M. S. and Haspel-math, M., editors, The World Atlas of Language Structures Online. Max Planck DigitalLibrary, Munich.

Dryer, M. S. (2011f). Order of numeral and noun. In Dryer, M. S. and Haspelmath, M.,editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

Dryer, M. S. (2011g). Order of Subject, Object and Verb. In Dryer, M. S. and Haspelmath,M., editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

Dryer, M. S. (2011h). Polar questions. In Dryer, M. S. and Haspelmath, M., editors, TheWorld Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Dryer, M. S. (2011i). Position of negative morpheme with respect to subject, object,and verb. In Dryer, M. S. and Haspelmath, M., editors, The World Atlas of LanguageStructures Online. Max Planck Digital Library, Munich.

67

Dryer, M. S. (2011j). Position of polar question particles. In Dryer, M. S. and Haspelmath,M., editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

Dryer, M. S. and Haspelmath, M., editors (2011). The World Atlas of Language StructuresOnline. Max Planck Digital Library, Munich.

Dupratz, P. (1864). Lettre du P. Dupratz. Annales de l’Oeuvre pontificale de la Sainte-Enfance, 6.

Echenberg, M. (1978). Tragedy at thiaroye: The senegalese soldiers’ uprising of 1944. InPeter Gutkind, R. C. and Copans, J., editors, African Labor History. Sage, London.

Echenberg, M. (1986). Slaves into soldiers: social origins of the tirailleurs senegalais. InCurtin, P. and Lovejoy, P., editors, Africans in bondage, pages 311–333. University ofWisconsin Press, Madison.

Fargeas, L. (1899). Notes sur la Côte d’Ivoire (suite). À travers le monde, 13.

Foley, W. (2006). Universal constraints and local conditions in Pidginization. Case studiesfrom New Guinea. Journal of Pidgin and Creole Languages, 21:1:1–44.

Gil, D. (2005). From Repetition to Reduplication in Riau Indonesian. In Hurch, B., editor,Studies on Reduplication. Mouton de Gruyter, Berlin.

Guillaume Balavoine (2013). Atlas-historique.net. Website [accessed on 2013-05-16].

Haegeman, L. (1997). Register variation, truncation, and subject omission in English andin French. English Language & Linguistics, 1.

Harris, M. (1988). French. In Harris, M. and Vincent, N., editors, The Romance Languages,pages 209–245. Croom Helm, London.

Harrison, A. (2003). Fulfulde language family report. SIL Electronic Survey Reports,2003-009.

Haspelmath, M. (2010a). Comparative concepts and descriptive categories and in cross-linguistic studies. Language, 86:663–687. Max Planck Institute for Evolutionary Anthro-pology.

Haspelmath, M. (2010b). Framework-free grammatical theory. In Heine, B. and Narrog,H., editors, The Oxford Hhandbook of Linguistics Analysis. Oxford University Press,Oxford.

Haspelmath, M. and the APiCS Consortium (2013). Expression of pronominal subjects.In Michaelis, S. M., Maurer, P., Haspelmath, M., and Huber, M., editors, Atlas ofPidgin and Creole Language Structures Online. Max Planck Institute for EvolutionaryAnthropology, Leipzig.

68

Holm, J. (1988). Pidgins and Creoles: Volume 1, Theory and structure. Cambridge Uni-versity Press, Cambridge.

Hopper, P. J. (1991). On some principles of grammaticalization. In Traugott, E. C.and Heine, B., editors, Approaches to Grammaticalization, volume 1. John Benjamins,Amsterdam.

Huber, M. and the APiCS Consortium (2013). Order of possessor and possessum. InMichaelis, S. M., Maurer, P., Haspelmath, M., and Huber, M., editors, Atlas of Pidginand Creole Language Structures Online. Max Planck Institute for Evolutionary Anthro-pology, Leipzig.

Hubin, G. (1987). Au fil de mes jours. ma vie - mes campagnes - ma guerre. volume 3.

König, E. and Gast, V. (2006). Focused assertion of identity: A typology of intensifiers.Linguistic Typology, 10.

Laverrière, S., editor (1932). Les Missions catholiques : bulletin hebdomadaire de l’Oeuvrede la propagation de la foi Lyon. Challamel, Paris.

Lefebvre, C. and Brousseau, A.-M. (2002). A grammar of Fongbe. Mouton grammar library.Mouton de Gruyter, Berlin.

Lewis, P. M., Simons, G. F., and Fennig, C. D., editors (2013). Ethnologue:Languages of the World, Seventeenth edition. SIL international. Online version:http://www.ethnologue.com/, Dallas, Texas.

Leymaire, H. (1898). La mort de Mabiala le féticheur. À travers le monde.

Lhote, H. (1947). Dans les campements touaregs. Les Œuvres Française, Paris.

Little, R. (2009). Lucie Cousturier, les tirailleurs sénégalais et la question coloniale.L’Harmattan, Paris.

Mangin, C. (1910). La Force Noire. À travers le monde 41. Hachette, Paris.

Maran, R. (1921). Batouala: véritable roman nègre. Albin Michel, Paris.

Marceau, H. V. (1911). Le tirailleur soudanais. Berger-Levrault, Paris & Nancy.

Marie-Victoria, S. (1921). Les petits Sénégalais à Biskra. Les Missions Catholiques 53.

Maurer, P. and the APiCS Consortium (2013). Gender agreement of adnominal adjectives.In Michaelis, S. M., Maurer, P., Haspelmath, M., and Huber, M., editors, Atlas ofPidgin and Creole Language Structures Online. Max Planck Institute for EvolutionaryAnthropology, Leipzig.

69

McGregor, W. B. (2009). Introduction. In McGregor, W. B., editor, The Expression ofPossession, volume 2 of The Expression of Cognitive Categories. Mouton de Gruyter,Berlin.

Michaelis, S. M., Maurer, P., Haspelmath, M., and Huber, M., editors (2013). APiCSOnline. Max Planck Institute for Evolutionary Anthropology, Leipzig.

Michel, M. (2003). Les Africains et la Grande Guerre. Kathala, Paris.

Miestamo, M. (2007). Negation – An Overview of Typological Research. Language andLinguistics Compass.

Montrozier, R. C. d. (1902). Deux ans chez les anthropophages et les sultans du centreafricain. Plon, Paris.

Muysken, P. and Smith, N. (1995). The study of pidgin and creole languages. InJacques Arends, Pieter Muysken, N. S., editor, Pidgins and Creoles. An Introduction.John Benjamins, Amsterdam, Philadelphia.

Nariyama, S. (2004). Subject ellipsis in english. Journal of Pragmatics, 36(2):237 – 264.

Nordenck, C. d. (1886). Voyage aux pays des bagas et du rio-nuñez. Le Tour du Monde,pages 273–304.

Nordhoff, S., Hammarström, H., and Haspelmath, M. (2012). Glottolog/Langdoc Online.

Parkvall, M. (2000). Out of Africa. Battlebridge, London.

Ruault, E. (2008). Un officier et la conquête coloniale. Presses Universitaires de Bordeaux,Bordeaux.

Rubino, C. (2011). Reduplication. In Dryer, M. S. and Haspelmath, M., editors, TheWorld Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Sembène, O. (1987). Camp de thiaroye (movie).

Sonolet, L. (1911). Les progrè de l’Afrique occidentale française 3: l’Organisation militaire,les troupes noires. Le Tour du Monde, pages 37–48.

Stassen, L. (2011a). Predicative adjectives. In Dryer, M. S. and Haspelmath, M., editors,The World Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Stassen, L. (2011b). Predicative possession. In Dryer, M. S. and Haspelmath, M., editors,The World Atlas of Language Structures Online. Max Planck Digital Library, Munich.

Stassen, L. (2011c). Zero copula for predicate nominals. In Dryer, M. S. and Haspelmath,M., editors, The World Atlas of Language Structures Online. Max Planck Digital Library,Munich.

70

Thomason, S., editor (1997). Contact Languages: A wider perspective. John Benjamins,Amsterdam.

Thornton, J. K. (1992). Africa and Africans in the making of the Atlantic world, 1400-1680. Cambridge University Press.

Van Den Avenne, C. (2012). Le petit manuel français-bambara à époque colonial, entredescription et appropriation pratique. Canadian Journal of African Studies/La Revuecanadienne des études africaines, 46.

Veselinova, L. (2010). Standard and special negators in the slavonic languages: Synchronyand diachrony. In Diachronic Syntax of the Slavonic Languages, pages 197–210. WienerSlawistischen Almanach.

Veselinova, L. (2013). Lexicalized negative verbs: a cross-linguistic study. In Associationfor Linguistic Typology 10th Biennial Conference (ALT 10).

Wetzer, H. (1996). Nouniness and Verbiness: a Typological Study of Adjectival Predication.Mouton De Gruyter, Berlin.

Wilson, M. (1999). Français-Tirailleur – The Pidgin French of France’s African Troops,1916. Student thesis, University of Auckland, Auckland.

Woodfork, J. (2001). Senegalese soldiers in the Second World War: Loyalty and identitypolitics in the French colonial army. PhD thesis, University of Texas at Austin.

Zipf, G. K. (1935). The Psychobiology of Language. Houghton-Mifflin, Oxford.

71